Fungal diseases: Health burden, neglectedness, and potential interventions

Editorial note

This report is a “shallow” investigation, as described here, and was commissioned by Open Philanthropy and produced by Rethink Priorities from January to February 2023. We revised the report for publication. Open Philanthropy does not necessarily endorse our conclusions, nor do the organizations represented by those who were interviewed.

Our report focuses on exploring fungal diseases as a potential new cause area for Open Philanthropy. We assessed the current and future health burden of fungal diseases, provided an overview of current interventions and the main gaps and barriers to address the burden, and discussed some plausible options for philanthropic spending. We reviewed the scientific and gray literature and spoke with five experts.

While revising the report for publication, we learned of a new global burden study (Denning et al., 2024) whose results show an annual incidence of 6.5 million invasive fungal infections, and 3.8 million total deaths from fungal diseases (2.5 million of which are “directly attributable” to fungal diseases). The study’s results align with this report’s estimate of annual 1.5 million to 4.6 million deaths (80% confidence) but were not considered in this report.

We don’t intend this report to be Rethink Priorities’ final word on fungal diseases. We have tried to flag major sources of uncertainty in the report and are open to revising our views based on new information or further research.

Executive summary

While fungal diseases are very common and mostly mild, some forms are life-threatening and predominantly affect low- and middle-income countries (LMICs). The evidence base on the global fungal disease burden is poor, and estimates are mostly based on extrapolations from the few available studies. Yet, all experts we talked to agree that current burden estimates (usually stated as >1.7M deaths/year) likely underestimate the true burden. Overall, we think the annual death burden could be 1.5M - 4.6M (80% CI), which would exceed malaria and HIV/AIDS deaths combined.¹ Moreover, our best guess is that fungal diseases cause 8M - 49M DALYs (80% CI) per year, but this is based on our own back-of-the-envelope calculation of high-uncertainty inputs.

Every expert we spoke with expects the burden to increase substantially in the future, though no formal estimates exist. We project that deaths and DALYs could grow to approximately 2-3 times the current burden until 2040, though this is highly uncertain. This will likely be partly due to a rise in antifungal resistance, which is especially problematic as few treatment classes exist and many fungal diseases are highly lethal without treatment.

We estimate that only two diseases (chronic pulmonary aspergillosis [CPA] and candidemia/invasive candidiasis [IC/C]) account for ~39%-45% of the total death and DALY burden. Moreover, a single fungal pathogen (Aspergillus fumigatus) accounts for ~50% of the burden. Thus, much of the burden can be reduced by focusing on only a few of the fungal diseases or on a few pathogens. Available estimates suggest the top fungal diseases have highest burdens in Asia and LMICs, and that they most affect immunocompromised individuals.

Fungal diseases seem very neglected in all areas we considered (research/R&D, advocacy/lobbying, philanthropic spending, and policy interventions) and receive little attention even in comparison to other diseases which predominantly affect LMICs. For example, we estimate the research funding/death ratio for malaria to be roughly 20 times higher than for fungal diseases. Moreover, fewer than 10 countries have national surveillance systems for fungal infections, and the WHO has no fungal disease program yet. Some fungal diseases are limited to certain geographical areas and are thus not considered a global priority by the WHO, yet they have a significant national burden (e.g., paracoccidioidomycosis is endemic in Brazil and has an incidence of 40 cases per 100k inhabitants in some areas).

Major barriers exist at every stage of the care cascade. In our (low-confidence) view, the top three obstacles to tractability are:

1. A limited availability (due to both lack of deployment and development) of accurate diagnostics that are inexpensive and easy to use hampers diagnosis, surveillance, and treatment efforts.
2. R&D costs for the development of vaccines and treatments are high (e.g., as antifungal treatments are commonly toxic for humans, finding both effective and non-toxic candidates is harder and more costly) and market incentives are insufficient.
3. The dual use of antifungals in clinical and agricultural contexts is unregulated and might be a key driver of antifungal resistance.

Several experts pointed out potentially promising interventions, such as:

• Funding R&D efforts to develop antifungal vaccines and new treatments.
• Supporting diagnostic efforts by, e.g., targeted screening for HIV patients.
• Improving surveillance by advocating for, e.g., the establishment of several sentinel surveillance systems globally and the inclusion of fungal diseases in the GBD study.
• Supporting antifungal stewardship by, e.g., banning some antifungals from use in agriculture.

We have not investigated these interventions in detail and have a very limited sense of how promising and tractable they are, whether they are suited for philanthropic funding, and how they could be supported more concretely.

We estimated the social return-on-investment of directly providing treatment for the top two fungal diseases in terms of health burden in LMICs and came away rather pessimistic. We found the barriers to treatment to be high for both CPA and IC/C (e.g., high side effects, long treatment duration, difficult diagnosis, high cost of treatment). Our best guess is that a life could be saved for at least $630 for CPA and at least $2.6k for IC/C in our most optimistic scenario, but this doesn’t take into account costs for diagnosis and hospitalization.

Current burden: Fungal diseases cause 1.5M - 4.6M deaths (80% CI) and 8M - 49M DALYs (80% CI) annually, with the burden concentrated among a handful of diseases, and most affect LMICs and immunocompromised people

We provide background information on fungal diseases in Appendix A, including:

• How are fungal diseases categorized?
• What causes fungal diseases, and who is most at risk from them?
• How are fungal diseases diagnosed?
• How are fungal diseases treated?
• What is antifungal resistance?

Below, we consider the burden in three main ways. Firstly, we consider the total death and DALY burden; secondly, we estimate how that burden is distributed among different diseases; and thirdly, we look at the geographical and comorbidity distribution of the burden.

Total burden: We estimate that the current annual global burden of all fungal diseases is 1.5M - 4.6M deaths (80% CI) and 8M - 49M DALYs (80% CI)

[Confidence: Overall 80%, as encapsulated in confidence intervals. Medium-high confidence that literature estimates are too low based on multiple expert interviews. Our estimate of the annual global burden has wide confidence intervals due to combined uncertainty around prevalence/incidence, mortality rates, counterfactual life expectancies, disability weights, and duration of illness.]

Data on the current global burden of fungal diseases is scarce, decentralized, and of generally low quality. DALY estimates do not exist for almost all fungal diseases:

• A few global incidence, prevalence, and mortality statistics are available (e.g., Bongomin et al., 2017; Brown et al., 2012; Fungal Infection Trust [FIT], 2020; Global Action for Fungal Infections [GAFFI], 2016),² but these are generally sourced from individual, non-systematic studies, or extrapolated from countries in which data are available. Our impression is that the global estimates were done in a fairly quick, back-of-the-envelope way.
• Moreover, every expert we spoke with agreed that these estimates likely underestimate the true burden due to underdiagnosis.³
• Our overall impression is that uncertainty is highest for diseases which likely have the highest death burden, possibly because these mostly occur in LMICs where diagnostics and surveillance is even lower than in HICs.
• The only currently existing global DALY estimates for fungal diseases are focused on fungal skin diseases only — which are by and large non-lethal — in the Global Burden of Disease Study (IHME, 2020a).⁴

We estimate that, annually, there are 1.5M - 4.6M (80% CI) fungal disease deaths (point estimate: 3M) and 8M - 49M (80% CI) DALYs (point estimate: 27M):

• We compiled these estimates in three stages (see Appendix B for a summary of our methodology, assumptions and sources. Our calculations can be found here):
  • Stage 1: We did a first, quick estimation of the burden for 30 fungal diseases. We took death estimates from the literature and adjusted them for underdiagnosis. We then calculated DALYs using rough assumptions.
  • Stage 2: Given significant uncertainty around burden estimates from the literature, we spent more time refining our estimates for the top two fungal diseases from stage 1, i.e., chronic pulmonary aspergillosis and invasive candidiasis/candidemia. We first refined our death estimates.
  • Stage 3: Finally, we refined our years of life lost (YLL) estimates for chronic pulmonary aspergillosis and invasive candidiasis/candidemia.
• Figure 1 shows an overview of our incidence, death, and DALY estimates for the top 13 fungal diseases (ranked by DALYs). A more comprehensive overview of the estimated burden and affected populations for each fungal disease can be found in a large table here.

Burden by disease: Only two fungal diseases are responsible for ~39-45% of the global death and DALY burden, though our ranking of diseases is highly uncertain

[Confidence: Low-medium confidence regarding death estimates and extremely low confidence regarding DALY estimates given heavy use of assumed life expectancy values (see Appendix B). Low confidence about the rank ordering of any two given fungal diseases and medium confidence that the burden is highly concentrated among a handful of fungal diseases. According to David Denning, some but not all fungal diseases might be so severely underdiagnosed that their true number of cases might be several times higher (up to 10 times higher for some diseases⁵).]

There is considerable uncertainty regarding the disease burden of individual fungal diseases.⁶ Nonetheless, our analysis suggests that — excluding skin, hair, and nail fungal infections, which are rarely fatal — the top two diseases account for 39%-45% of the total burden (~39% of DALYs; ~45% of deaths) and the top six diseases account for ~67-84% of the total burden of fungal diseases (~67% of DALYs; ~84% of deaths; see Figure 1). In descending order of DALY burden, these six diseases are:

1. Chronic pulmonary aspergillosis (CPA): a chronic lung (i.e., systemic) infection most often caused by Aspergillus fumigatus; people with pre-existing lung conditions such as tuberculosis, COPD, and lung cancer are most at risk (Leading International Fungal Education [LIFE] - CPA).
2. Candidemia and invasive candidiasis (IC/C): an invasive bloodstream (i.e., systemic) infection most often caused by Candida albicans; people with immune deficiencies, including HIV/AIDS patients, premature infants, those undergoing chemotherapy, and those in intensive care units, are most at risk (LIFE - Invasive Candida).
3. Invasive aspergillosis (IA): an invasive lung (i.e., systemic) infection most often caused by Aspergillus fumigatus; people who receive bone marrow or lung transplants and leukemia patients are most at risk (LIFE - Invasive Aspergillosis & Mucormycosis).
4. Progressive disseminated histoplasmosis (PDH): an invasive bloodstream (i.e., systemic) infection caused by Histoplasma capsulatum; people with HIV/AIDS, those who are otherwise immunocompromised, and those at extremes of age are most at risk (LIFE - Disseminated Histoplasmosis).
5. Pneumocystis pneumonia (PCP): a subacute or rapidly progressive invasive lung (i.e., systemic) infection caused by Pneumocystis jirovecii; people with HIV/AIDS and other immunodeficiencies, including transplant recipients, leukemia patients, and malnourished children, are most at risk (LIFE - Pneumocystis Pneumonia).
6. Severe asthma with fungal sensitization (SAFS): (systemic) allergy to fungi such as Aspergillus fumigatus, Penicillium chrysogenum, and Cladosporium herbarum, which affects 1.75%-10% of all asthma patients; no specific risk factors other than asthma have been identified (LIFE - Fungal Asthma).

Note that we did this analysis at the disease level. Another possible option is to estimate the disease burden at the fungal pathogen level.⁷ However, we have not found an overview of the burden by pathogen. Moreover, as multiple fungal pathogens are known to cause the same disease, we expect that estimates of global burden at the pathogen level would be even less precise. Our rough guess is that Aspergillus fumigatus is the top fungal pathogen, causing ~50% of the DALY burden. This is because it causes many of the cases of CPA, IA, SAFS, and some other (smaller) fungal diseases. Note that aspergillosis infections can also be caused by other Aspergillus species, such as A. flavus, A. niger, and A. terreus (CDC, 2021). We would also like to point out that treatments are similar for different diseases caused by the same fungus. For example, azoles are the recommended first-line treatment for several aspergillosis infections, though the exact recommendation still varies somewhat across different diseases from the same pathogen (see here for more detail on how fungal diseases are treated).

Figure 1: A few diseases likely dominate the global death and DALY burden of fungal diseases

Note. Blue bars represent deaths and gray bars represent DALYs. Central estimates reported are midpoints of 80% confidence intervals, except for “skin, hair, and nail fungal infections,” where we used the 2019 Global Burden of Disease estimate. Data labels (percentages) refer to proportion of total global fungal disease burden. Estimated by Rethink Priorities based on literature sources, expert opinion, and subjective analysis. Full calculations here, with further explanation of the methodology in Appendix B.

Distribution of burden: Fungal diseases with the highest burden are most common in LMICs and among immunocompromised individuals

[Confidence: Low confidence on the continental and World Bank income classification distributions due to scarce data that is likely skewed by differences in diagnostic efforts. We have medium confidence on the most susceptible population groups, but have not found any comprehensive statistics for all fungal diseases. We have low confidence on the most affected age groups, as we found only little information in a quick search.]

Current estimates point to the top fungal diseases (with the highest death and/or DALY burden) being most common in LMICs, particularly in Asia,¹ though we have low confidence in those estimates:

• No comprehensive estimates of the distribution of fungal diseases across countries or regions exist. Global Action for Fungal Infections (GAFFI) has a collection of prevalence/incidence maps for some fungal diseases, but we believe that these are highly skewed by differences in diagnostic efforts.² Thus, we think these maps should be taken very lightly.
• If we (somewhat) believe the available data, two observations point to the number of cases — and by extension deaths/DALYs — being potentially highest in LMICs (particularly in Asia):
  • The top two diseases in terms of deaths and DALYs,³ i.e. CPA and IC/C, are reported to have the highest incidence/prevalence rates in LMICs (see Appendix C).
  • In Appendix D, we present an analysis of the available country-level estimates, primarily sourced from Bongomin et al. (2017). For the highest burden fungal diseases, we show the rough burden distributions by World Bank income classification (Figure E1) and by continent (Figure E2). In summary, we find that available country-level data — which again could be highly skewed by differences in diagnostic efforts — suggest:
    • Most of the major fungal diseases appear to have the highest case numbers in Asia (~78%, ~66%, and ~93% cases of the top three diseases, respectively), except for Pneumocystis pneumonia, of which most cases are in Africa (~61%).
    • All the major fungal diseases appear to have the highest case numbers in LMICs (~99%, ~88%, and ~97% cases of the top three diseases, respectively), although severe asthma with fungal sensitization has disproportionately high prevalence in HICs.

Immunocompromised individuals are more susceptible to fungal diseases and are also more likely to experience complications and death:

• We tried to get a sense of how three of the top fungal diseases are distributed across populations with different underlying conditions and found that:
  • Chronic pulmonary aspergillosis almost always occurs in patients with respiratory diseases:
    • ~50% (range 20%-85%) of cases occur in tuberculosis patients (Bongomin et al., 2017, p. 11). Our guess is that ~40% of cases are in patients with other respiratory diseases (e.g., sarcoidosis, COPD)⁴, and the remaining ~10% in healthy populations.
  • Invasive candidiasis and candidemia almost always occurs in immunocompromised and/or hospitalized patients:
    • ~60% of detected cases “were reported in the ICU followed by cancer and transplant units (13%)” (Bongomin et al., 2017, p. 4).⁵ Our guess is that ~20% of cases occur among various other underlying conditions (e.g., diabetes, use of broad-spectrum antibiotics), and ~7% among healthy populations.
  • Progressive disseminated histoplasmosis usually occurs in patients with various underlying diseases, but mostly HIV/AIDS:
    • Our rough guess is that ~35% of cases occur among patients with HIV/AIDS, ~30% among various other underlying disease groups (e.g., respiratory diseases), and the remaining ~35% among healthy populations (FIT, 2020, p. 8).

For the vast majority of fungal diseases, we have not been able to find any distribution of cases across age groups. Thus, our (low-confidence) guess is that most fungal diseases affect all age groups:

• An exception is progressive disseminated histoplasmosis, for which being in the “extremes of age” is a risk factor (see Appendix C).
• David Denning (professor of infectious diseases in global health, University of Manchester) described aging as a risk factor for fungal diseases (see also Jenks et al., 2023), though our overall impression from reading the literature is that fungal diseases are much more determined by underlying conditions (such as HIV/AIDS) than by age per se.
• We spent little time on this question (~1h) and it’s possible that more hours of research could provide a clearer picture on the age distribution.

Future burden: Based on a quick calculation, our best guess is that fungal disease deaths and DALYs could grow to approximately 2-3x the current burden by 2040

[Confidence: Low. We are fairly confident that the number of deaths will increase (as stated by several experts). Nonetheless, in the absence of existing modeling efforts and due to several highly unpredictable factors, our future burden estimate is merely speculation.]

Experts agree that the future burden of fungal diseases is most likely to grow, but no modeling studies exist:

• No modeling studies of the future health burden of fungal diseases exist and no expert we spoke with was willing to provide a best-guess quantitative estimate.⁶
• We have seen a commentary that discusses the potential for growth in qualitative terms and concludes that “[a]ll trends suggest that the importance of fungal diseases will increase in the 21st century” (Casadevall, 2018). All experts we spoke with agree with this assessment.

A rise in predictable risk factors is likely to cause a rise in fungal disease incidence and mortality, but several highly unpredictable factors could potentially dominate:

• From conversations with several experts, we learned that the future incidence of fungal disease can broadly be considered a function of two components:
  • Somewhat predictable factors:
    • An increase in risk factors, such as cancer, HIV/AIDS, organ transplants, chronic respiratory diseases, tuberculosis, diabetes, aging, and intensive care patients.
  • Highly unpredictable factors: e.g.,
    • Climate change selecting for temperature-resistant fungi.
    • Emergence of a new viral disease like HIV that weakens immunity.
    • Emergence of new fungal pathogens and disease outbreaks.⁷
• The future mortality (and as a consequence, DALYs) of fungal diseases depends strongly on antifungal resistance:
  • According to Fisher, it is likely that there is much more widespread resistance among the most important fungal pathogens to frontline azole-based drugs by 2040. He believes that this could plausibly cause a 50%-100% increase in mortality. We are uncertain about this, as we have not been able to obtain any other estimates on the resistance timelines based on expert interviews or the literature.
• On the other hand, progress in R&D might reduce the future burden (e.g., via new drugs, diagnostics, vaccines, and immunotherapies; Casadevall, 2018). Given low current R&D funding (as we explain here), we expect the progress to be slow.

Based on a quick back-of-the-envelope calculation, we find it plausible that fungal diseases could grow to ~2-3x the current burden by 2040. This equates to 3.7M - 12.6M deaths and ~22M -138M DALYs (80% CI) in 2040. However, this remains highly uncertain:

• These figures are based on a simple model in Causal, which is based on the following assumptions:
  • Fungal disease deaths/DALYs in 2022: 1.5M - 4.6M deaths (80% CI) and 8M - 49M DALYs (80% CI) in 2022 (as explained here)
  • Annual population growth rate: 0.8%
  • Incidence growth rate due to predictable risk factors: 0.2%-1.2% (80% CI):
    • See Appendix E for an explanation. Note that our model only considers the incidence of risk factors, but not changes in their respective treatment. We expect that there will be countervailing effects due to an improved treatment for these risk factors in the future, which would dampen the relationship between incidence of risk factors and incidence of fungal diseases, but we did not have time to consider the likely extent of this.
  • Mortality growth rate due to antifungal resistance: 1%-3% (80% CI)
    • According to Fisher, mortality due to an increase in antifungal resistance could rise by 50%-100% by 2040. Based on this estimate, we back-calculated the compound annual growth rate, which is 2%-4%. We revised it slightly downwards to remain conservative.
  • Growth rate due to other, highly unpredictable factors: 1%-3% (80% CI):
    • According to Marcio Rodrigues (senior investigator at the Oswaldo Cruz Foundation), these unpredictable factors could potentially outweigh the other factors and be a key driver of growth. Nonetheless, we opted for a (in our view) conservative parameter.
  • We assume that deaths and DALYs grow at an equal pace. This may not be realistic.

Neglectedness: Fungal diseases appear to be very neglected in terms of research/R&D, advocacy/lobbying, philanthropic spending, and policy interventions

Research/R&D: We estimate that annual research funding for fungal diseases is less than USD 50M and annual commercial R&D is USD 70M - 150M (70% CI)

[Confidence: Medium. All sources we’ve seen so far point to low R&D research spending and to a neglect of fungal disease research relative to other infectious diseases. However, we spent only a few hours on this and it’s possible that some sources are outdated and we missed some newer funding streams/research initiatives.]

Fungal diseases research funding is very low compared to research funding for other infectious diseases with similar mortality. We think it’s unlikely more than USD ~50M:

• According to Rodrigues, Aspergillus and Candida are most likely the top funded pathogens. Cryptococcus is in the middle, and neglected pathogens such as Paracoccidioides, Sporothrix, Fonsecaea, and many others are so poorly funded that they don’t even appear under estimates of research funding for neglected diseases (e.g., Chapman et al., 2021, p. 13).
• According to the G-FINDER project, 2020 (non-commercial) R&D funding for fungal diseases was USD 11.7M (this only includes cryptococcal meningitis, mycetoma, and histoplasmosis), which corresponds to 0.3% of total R&D funding for neglected diseases (Chapman et al., 2021, p. 13).⁸
• Given the expert view that Aspergillus and Candida are likely the top funded pathogens, and the fact that they are not included in the G-FINDER report, our very rough guess is that these are funded with ~USD 35M.⁹
• An analysis of UK research funding between 1997-2010 found 171 studies related to mycology (~12 studies/year), and a mean annual funding of GBP 3.5M (USD ~4.3M), ~90% of which was spent on preclinical work (Head et al., 2013).
• Research funding per DALY and per death is much lower for fungal diseases than for other diseases, such as HIV/AIDS, tuberculosis, and malaria (see Table 1).¹⁰
• 1.4%-2.5% of research funding on immunology and infectious diseases from Wellcome Trust, the UK Medical Research Council (MRC), and the US National Institutes of Health (NIH) was spent on mycology (Brown et al., 2012).

Table 1: Rough comparison of research funding (in USD) to DALY and death ratios in 2019 across diseases

Disease	Research funding/DALY ratio	Research funding/death ratio
Fungal diseases	4.4-7.4	40-67
HIV/AIDS	31.3	1736
Tuberculosis	15.3	608
Malaria	13.8	998

Note. We calculated these ratios based on 2019 global DALY and death estimates obtained from the GBD Results Tool and 2019 research funding figures from Chapman et al. (2021, p. 13). For fungal diseases, we assumed USD 120M - 200M of total research funding (USD 50M for public/philanthropic research + USD 70M - 150M for commercial R&D) and the midpoints of our own DALY/death estimates. We intend these figures to be illustrative rather than definitive.

Relatively few academic groups are focused on research related to fungal diseases:

• We provide some examples of eight major research groups in Appendix F.

R&D from commercial biotech companies seems to be limited but growing, with a best guess spending of around USD ~70M - 150M (70% CI). We think that only a handful of antifungal treatments, diagnostic tests, and vaccines are currently in the development pipeline:

• According to a 2022 market analysis by Emergen Research, R&D is growing.
• R&D by large pharma companies (e.g., Novartis, Bayer) seems very limited,¹⁸ with the exception of Pfizer, which recently acquired an antifungal biotech company.
• Our impression is that most of the current R&D is done by small to medium-sized biotech companies who each have about one to two antifungal or diagnostic tools in their pipelines. We don’t have a good sense of the total funding, but we expect that it’s somewhat higher than the total funding of examples shown in Appendix H.

Progress in R&D appears slow, but new fungal treatments and vaccines are currently in development:

• Few antifungal drugs are coming to market, and research has stalled since the 1990s.¹⁹ However, the US Biomedical Advanced Research and Development Authority (BARDA) recently added antifungal medical countermeasures to its areas of interest (Contract Pharma, 2022), indicating a readiness to fund antifungal R&D projects in partnership with the private sector.²⁰
• Antifungal drug development is costly and takes a long time as it’s hard to find candidate treatments that are effective yet well-tolerated by humans:²¹
  • The process from discovery to FDA approval takes more than 20 years and can cost USD ~1.7B per molecule according to Rex (2022).²²
• Only three fungal vaccines have ever reached clinical trials, none of which targets the diseases with the highest burdens or has been found to protect immunocompromised people (Del Poeta, 2021, 11:13-12:31; Oliveira et al., 2021):
  • One vaccine for coccidioidomycosis (Valley Fever)²³ and two vaccines for vulvovaginal thrush (vulvovaginal candidiasis).²⁴
• Several pan-fungal²⁵ vaccine candidates have been developed and tested in murine models, but only one candidate appears to have been significantly protective in non-human primates.²⁶

Advocacy/lobbying: We estimate that annual advocacy and lobbying expenditures are approximately USD 500k

[Confidence: Medium. We spent only ~4h on this and it’s likely that we missed some initiatives, but we expect these are likely to be much smaller in terms of budget than the ones mentioned here.]

Advocacy efforts concerned with fungal diseases appear rather limited and unlikely to have a total budget significantly higher than USD 500k/year:

• We found two major advocacy initiatives working on fungal diseases (GAFFI and Leading International Fungal Education [LIFE]), and believe that their total annual budget is likely around USD 500k (see Table 2).²⁷
  • These initiatives are mainly concerned with collecting data, raising awareness, and educating policymakers and medical professionals on fungal diseases.
• Some organizations (GAFFI, US CDC, MSF, CHAI) claim to have successfully lobbied jointly for the inclusion of two antifungals on the WHO Essential Medicines List, which as of 2021 includes eight antifungals (GAFFI, 2016; WHO, 2021).

Table 2: Main advocacy initiatives/programs related to fungal diseases

Initiative or program	Annual budget	Key areas of action
Global Action Fund for Fungal Infections (GAFFI)	~USD 370,00028	International organization focused on raising awareness and collecting global data on fungal disease Advocates for making fungal diagnostic tests and antifungal treatments available to 95% of global population by 2025
Leading International Fungal Education (LIFE)	~USD 70,000, funded by Fungal Infection Trust	Education organization, mainly aimed at medical community Participation in awareness campaigns and public engagement activities

Philanthropic spending: We estimate that annual philanthropic spending is less than USD 10M

[Confidence: Medium. We spent only a few hours on this and we expect that we haven’t found all the philanthropic funders that focus on fungal diseases. However, we’re fairly confident that our spending estimate is not off by an order of magnitude.]

Our impression is that most philanthropic spending on fungal diseases comes from three initiatives (see Table 3) with a total budget of around USD 10M or lower per year:

• Philanthropic spending is partly spent on supporting research, and partly on some of the advocacy efforts outlined here.
• Other small initiatives exist (e.g., Fondation JYLAG, Fondation Leenards, BIO Ventures for Global Health, W.M. Keck Foundation).

Table 3: Main philanthropic funders in fungal disease research and education

Name	Annual philanthropic spending/grants	Key areas of action
Wellcome Trust	Unclear, but likely University of Aberdeen in 2012; USD 300,000 in 2020 for cryptococcal meningitis [Chapman et al., 2021, pp. 42-43])	Likely the largest philanthropic funder for fungal disease research, but fungal diseases are not mentioned as a focus area and are hardly mentioned on their website
Fungal Infection Trust (FIT)	Average annual grants of ~USD 170,000 Less than USD 50,000 for research grants in 2020	Aims to promote education about fungal diseases, support scientific research, improve diagnosis and treatment, and support the training of professionals
Burroughs Wellcome Fund	Have not found any figures, but most likely below USD 1M, probably much less	Seem mainly focused on supporting academic research and advocacy (e.g., GAFFI) on fungal diseases, but to a limited extent

Policy interventions: Fungal diseases are highly neglected by governments and international bodies

[Confidence: Medium. We spent only a few hours of research on this, but experts confirmed our impression that government policy/health systems interventions are very limited.]

Fungal diseases receive very little attention in government policy/health-care systems interventions at the national level:

• Note that most publicly accessible information we found on this is topic from 2015 or 2017. Our general impression from talking to multiple experts is that this is largely still true, though we expect that the exact figures have likely changed over time.
• There are limited national surveillance systems for fungal infections:
  • A 2017 editorial in Nature Microbiology reported that, “fewer than 10 countries have national surveillance programs for fungal infections, and fewer than 20 have fungal reference diagnostic laboratories” (“Stop neglecting fungi”, 2017).
  • Tom Chiller (director of the CDC Mycotic Diseases Branch) told us that fungal disease surveillance in the US is limited to coccidioidomycosis and Candida auris.
• Few countries have comprehensive education and training programs related to fungal diseases:
  • According to GAFFI (2015, Appendix F), only two master’s programs focusing on fungal diseases existed worldwide (Universities of London and Manchester) as of 2015.²⁹ Moreover, “very little is taught in public health and global health courses about fungal diseases” (GAFFI, 2015, Appendix F).
• In 2015, no national clinical guidelines or concrete public health strategies existed for several fungal diseases, e.g., mycetoma, fungal keratitis, and fungal asthma (GAFFI, 2015, p. 14). We have not reviewed to what extent this is still the case.
• There was “no recognized national or international authority on public health mycology” in 2015 (GAFFI, 2015, p. 15). Our impression is that this is still true.

At intergovernmental bodies like the World Health Organization, fungal diseases also receive limited attention. Currently, there is no WHO program on fungal diseases (besides GLASS, which is only concerned with antimicrobial resistance³⁰).

• Chiller said that the CDC and WHO have a cooperative funding agreement whereby the CDC provides the WHO with USD 150,000 per year, and that the CDC accounts for the vast majority of funding for the WHO’s fungal activities.
  • The present five-year contract will lapse in 2024, and upon renewal it is possible that the CDC may double its support to USD 300,000 per year.
  • According to Chiller, the WHO has limited fungal expertise relative to the organization’s overall scale. The US CDC has been involved with efforts to bolster the WHO’s capacity on fungal diseases, including potentially establishing a fungal diseases unit.
• A major recent initiative is the WHO (2022) Fungal Priority Pathogens list (FPPL), “the first global effort to systematically prioritize fungal pathogens” that also offers a guide for public health action.
  • According to Denning, the WHO is planning to start actions soon (at the time of our interview, GAFFI and WHO were in talks about this). These actions will be broadly focused on promoting surveillance activities and fostering dialogue around fungal diseases, but we are unaware of further details.
  • According to Rodrigues, a possible consequence of the WHO FPPL is that the most neglected fungal diseases will likely be even more neglected, as these are not on the FPPL because they are limited to some countries/regions.³¹ He mentioned his upcoming publication (now published; Rodrigues, 2023), a critical commentary on the FPPL.

Tractability and interventions: Overcoming diagnostic limitations, agricultural-clinical dual use, and commercial barriers to R&D are critical for tackling fungal diseases

[Confidence: Low-medium overall. This section is mostly a 3.5-day synthesis of literature reviews, textbooks, and expert opinion with only minor first-principles input from our own analysis. With limited historical philanthropic interventions to review, it is difficult to identify specific points of failure and opportunities.]

Tractability concerns by stage of care: Major barriers exist at every stage of tackling fungal diseases

[Confidence: Medium.]

Below, we present some of the major barriers to tackling fungal diseases and their implications:

• Prevention:
  • Barriers: Scientific, logistical, and commercial barriers have, to date, prevented the successful development and deployment of fungal vaccines (Oliveira et al., 2021). Alternative prevention options are limited, as fungi are ubiquitous, most life-threatening fungal infections are opportunistic, and the growth of the at-risk, immunocompromised population is hard to reverse.
  • Implications: Without effective vaccines or alternative prevention methods, advances in diagnosing and treating fungal diseases could be outpaced by the growth in at-risk population, which could be gradual or sudden.³²
• Diagnosis:
  • Barriers: Existing diagnostic methods are unavailable in many countries, partly due to lack of clinical suspicion³³ and test unavailability (GAFFI, n.d.-a). Scientific challenges mean that certain fungal diseases are — even in well-resourced settings — poorly diagnosed, with the “gold standard” of culture and microscopic techniques often highly insensitive (Lass-Flörl, 2017, pp. 3-15).
  • Implications: A lack of cheap and effective diagnostics could critically thwart the effective implementation of any advances in treatment and surveillance, which are downstream of diagnosis.
• Treatment:
  • Barriers: Effective current-generation treatments are inaccessible or unaffordable in many developing countries,³⁴ and in LMICs most hospitals and health-care workers have not been trained to administer certain treatments (GAFFI, n.d.-b; GAFFI, 2014). The development of next-generation treatments faces significant scientific, logistical, and commercial challenges (Perfect, 2017).
  • Implications: A lack of cheap and effective treatments would be directly responsible for poor prognoses and high case fatality rates. New treatments need to be developed and made available in LMICs in order to keep apace with resistance.
• Surveillance:
  • Barriers: The WHO’s recent recognition of fungal diseases as a public health problem has yet to trickle down to the local government level, according to Rodrigues. Most international and national public health initiatives do not collect data that would allow for the tracking of fungal diseases’ global burden (Bongomin et al., 2017). Antimicrobial resistance surveillance studies largely exclude fungal diseases, meaning there will likely be insufficient data on resistance for the foreseeable future (Fisher et al., 2022).
  • Implications: Insufficient epidemiological data could make advocacy for public health action less convincing and thus, per dollar, more difficult. Inadequate resistance surveillance could misdirect treatment R&D priorities.
• Antifungal stewardship:
  • Barriers: Underregulated agricultural use of fungicides drives clinical drug resistance, especially for azoles (Fisher et al., 2018). Current susceptibility testing is inadequate, partly due to intrinsic scientific limitations and partly due to different testing practices in Europe and the US (Fisher et al., 2022).
  • Implications: Failures in stewardship could mean that resistance spreads or strengthens faster than new treatments are developed.

Top tractability concerns: The top three barriers are likely diagnostic limitations, agricultural-clinical dual use of antifungals, and commercial barriers to drug and vaccine R&D

[Confidence: Low-medium. Our ranking is informed by evidence review and expert interview, but we haven't run our ranking past experts and we think it's likely tractability could be affected by known unknowns (variation by disease characteristics, population, context, etc.) and unknown unknowns.]

Top structural barrier to effective diagnosis (and treatment and surveillance): Highly specific and sensitive, easy-to-administer, and inexpensive diagnostic tests have yet to be deployed and, in some cases, developed.

• Note that here we condense three stages of the care cascade into one. Our sense from speaking with several experts is that the top diagnostic barriers are by extension those of effective treatment and surveillance.³⁵
• Conditional on failure to deploy and develop such tests, the marginal value of funding surveillance programs and treatment deployment would be lower. The ability of surveillance programs to detect disease and the ability of therapeutics to reach those who need them would be significantly compromised.

Top structural barrier to effective antifungal stewardship: The dual use of antifungals and fungicides in clinical and agricultural settings, which drives and accelerates resistance, is unregulated (Fisher et al., 2018; UK Environment Agency, 2022).

• Fisher indicated in our interview that government agencies’ regulatory structures present a major obstacle to preventing dual use. For example, in the US, the EPA and FDA do not currently have channels by which to cooperatively monitor and restrict the use of antifungals. We infer that the problem likely also exists in other countries and could be hard to coordinate globally.
• Conditional on failure to regulate the agricultural-clinical dual use of antifungals, the marginal value of funding antifungal development would be lower. New generations of antifungals — if not ring-fenced for clinical use — could more quickly succumb to resistance.

Top structural barrier to effective prevention: Potentially limited revenues — combined with high R&D and trial costs — of potential antifungal drugs and fungal vaccines could prevent their commercialization (Perfect, 2017; Oliveira et al., 2021).

• In interviews, Denning, Rodríguez Tudela, and Rodrigues singled out limited market incentives — especially for diseases that predominantly affect those in LMICs — as a major challenge for vaccine and drug development.³⁶
• Conditional on failure to rectify market incentives, the marginal value of funding vaccine and drug development could be lower. The same amount of philanthropic funding would yield fewer commercially available vaccines and drugs.

Interventions: Promising interventions include facilitating antifungal drug R&D and improving diagnostics in LMICs

[Confidence: Low re: our selection of plausibly best options; medium re: the list of interventions. We directly took these suggested interventions from a handful of expert interviews, but we did not have time to evaluate how tractable they are, whether they are suited for philanthropic funding/grantmaking, and how these efforts could be supported concretely. Moreover, some of the below suggestions come from advocacy organizations (e.g., GAFFI), who may tend to emphasize their own programs.]

With low confidence we recommend that, among the options below, an interested funder pay particular attention to the development of antifungals/vaccines and GAFFI’s diagnostic facilitation model.

First, there may exist high-leverage opportunities in the development of next-generation antifungals. For example, a funder could consider speaking with US NIAID and BARDA about their interest in funding antifungal drugs and fungal vaccines (Contract Pharma, 2022; NIAID, 2022), and consider funding opportunities that may not reach the bar for the US government but are nevertheless promising. There may be a role for philanthropists in particular in helping to overcome market-incentive barriers for fungal diseases that have substantial burdens in LMICs.

Second, the diagnostic facilitation model GAFFI has developed could be ready to scale soon, given the existing evidence of reduced mortality among a reasonably traceable and accessible population (HIV patients) and its continued piloting in Latin America.³⁷ As discussed, improvements in diagnostic infrastructure represent a critical entry point toward wider improvements in treatment and surveillance.

Here is a more complete list of interventions that were endorsed by experts we spoke with:

• Prevention:
  • Development of vaccines
  • Antigen screening programs that can detect the presence of pathogens before the onset of disease, such as that in Ramachandran et al. (2017)
• Diagnosis:
  • GAFFI’s diagnostics facilitation model (work in progress).³⁸ Its pilot program in Guatemala reportedly reduced mortality rate by 7 percentage points by screening HIV patients for fungal diseases. We reviewed the underlying study (Medina et al., 2021) and would put little weight on this specific finding.³⁹
  • Shift away from existing laboratory methods that require culturing microorganisms and toward fungal-specific tests (such as the cryptococcal antigen test): encompassing a range of specific interventions from educating physicians, distributing diagnostic tests, and developing next-generation tests
  • Layering fungal disease diagnostics upon existing (esp. COVID-19) infrastructure, physician education, lab technician training, and improvements in IT systems
• Treatment:
  • Development of new antifungals (together with vaccines)
  • Top-down programs to tackle the most neglected fungal diseases as decided by the WHO, in the style of those against malaria, smallpox, and other tropical diseases (e.g., involving support across medical personnel training and drug distribution)
  • Research investments for the most neglected fungal diseases (specifically, those whose aetiological agents did not make it into the WHO FPPL or its highest tiers)
• Surveillance:
  • Inclusion of fungal diseases in the Global Burden of Disease Study⁴⁰
  • Global digital database, which takes input from laboratories around the world, to estimate the health burden
  • Establishment of ~20 sentinel surveillance systems around the world with uniform diagnostic procedures
• Antifungal stewardship:
  • Ring-fencing of novel antifungals’ modes of action for clinical use, especially azoles for Aspergillus, in order to prevent agricultural use
  • Innovations in circular agriculture that mitigate the selection of resistance in crop pathogens
  • Efforts to identify and take action regarding ecological hotspots that promote high growth of fungi in the presence of fungicides, accelerating environmental resistance

We estimate the cost-effectiveness of providing treatment for the top two fungal diseases to be low

[Confidence: Low-medium. We are reasonably confident in our estimates for the treatment cost per case across several countries, but have low confidence in our estimated cost per life saved, as there is little reliable information regarding mortality and effectiveness of fungal disease treatments. We’re confident that the barriers to treatment are substantial for both CPA and IC/C (e.g., high side effects, required hospitalization for IC/C treatment, long treatment duration)].

In this section, we estimate the direct social return-on-investment (SROI) of providing treatment for the top two fungal diseases in terms of health burden in LMICs: chronic pulmonary aspergillosis and invasive candidiasis/candidemia. If cheap treatments exist which are effective and can be easily distributed and administered, there is a stronger case for focusing on fungal diseases as a cause area.

See Appendix G for more detail on our inputs, assumptions, and sources, and here for our calculations for several example countries.

• Chronic pulmonary aspergillosis:
  • We are pessimistic about the SROI of CPA treatment for several reasons: (1) treatment is relatively costly in most countries and usually requires daily medication of at least 6-12 months; (2) treatment effectiveness seems rather limited (many treated people relapse and mortality remains high) though data on effectiveness is pretty limited; (3) side effects are high; (4) diagnosis is difficult, even in rich countries.
  • Treatment costs vary tremendously across countries (cost per daily dose: ≥$0.13 in Vietnam; ≥$162 in China). Our best (highly uncertain) guess is that a life can be saved at a cost of at least ~$630 in Vietnam and ~$780k in China (in the most optimistic scenario), but we wouldn’t put much weight on those estimates. Moreover, this doesn’t take into account the cost of diagnosis, which can also be complicated and likely costly.
• Invasive candidiasis/candidemia:
  • We are confident that a treatment with echinocandins can be reasonably effective for IC/C, but we are not currently convinced that it is cost-effective. We estimate the medication cost per life saved to be at the very least $2.6k in our most optimistic scenario (high uncertainty), which would likely increase by 3x-4x if we include the cost of hospitalization. Moreover, current “gold standard” diagnostic tests have limited sensitivity and can take too long, which can increase mortality substantially. Treatment seems to be relatively straightforward to administer, but it needs to be given through the vein and requires a prolonged hospitalization.

Key uncertainties

Below we highlight the key uncertainties in our research on fungal diseases.

• We expect the death toll of fungal diseases to be higher than current estimates suggest, but we don’t know by how much. Due to poor-quality/lack of/inconvenient diagnostics and very few systematic surveillance efforts, many cases are missed. Moreover, several uncertainties in the evidence informing our model inputs could be affecting our estimates; we have tried to capture some of these using confidence intervals and conservative assumptions, but these remain key uncertainties, e.g.:
  • Incidence/prevalence: For some diseases, these statistics differ significantly across sources⁴¹ and/or were estimated in an intransparent way. For example, we tried to trace back the original source/estimation approach for the prevalence of chronic pulmonary aspergillosis, but got lost in a chain of citations that seemingly didn’t lead anywhere.
  • Mortality rate: We used the best available rates from two sources. However, it is possible that selection bias in diagnosis may have inflated the mortality rates (e.g., those with worse prognosis being more likely to be diagnosed and die), but we are unsure by how much. We revised estimated mortality rates downwards based on expert advice and evidence of recently decreased mortality rates in some fungal diseases. However, this assumption may have caused us to under- or overestimate mortality rate if this trend varies across fungal diseases (which we find plausible).
• We are very unsure about the DALY burden of fungal diseases. Due to comorbidities of fungal diseases, we are highly uncertain about our calculation assumptions:
  • YLLs: Fungal diseases differ in terms of comorbidities and comorbidities differ highly in how they affect individuals’ life expectancies. We are even more unsure for all diseases besides CPA and IC/C, as we did not have time to estimate YLLs for those separately.
  • YLDs: We are reasonably confident about our disability weight estimates (based on diseases with similar symptoms), but don’t have a good sense of the average duration of cases and expect these to differ widely across diseases.
• It’s possible that the recent publication of the WHO (2022) Fungal Priority Pathogens List will lead to significantly more international action related to fungal diseases and make the disease area less neglected. We learned in talks with experts that the WHO is currently planning some actions, but we don’t know what exactly, and whether and how governments are responding to the list.
• We spent the majority of the time on this report on our estimates of the burden, and only about five days on barriers, tractability, and potential interventions. Thus, we reviewed interventions and barriers only at a very superficial level and might have missed important considerations.

What we would do with more time

Burden:

• Review mortality burden for specific countries to identify fungal disease hotspots:
  • There seems to be a large heterogeneity in incidence across countries and this could help us identify if diseases are especially common in specific areas. We also expect the burden estimates to be more precise at the country level.
• Refine DALY burden estimates:
  • We used a lot of crude assumptions for our DALY estimates. For example, we assumed the same counterfactual life expectancy across almost all diseases and the same average duration of illness. We expect there to be a lot of heterogeneity across fungal diseases (as they differ in terms of their comorbidities) which our current model does not capture sufficiently. With more time, we could refine estimates for, e.g., our top six diseases.
• Conduct a more complete version of the country-level analysis outlined in Appendix D:
  • We did not have time to fully review the additional country-level burden studies and research posters/abstracts that have been published since — or were excluded from — Bongomin et al. (2017). With more time, we would continue to fill in the table here with data for more countries, for the sake of completeness, but this would likely have minimal effect on our conclusions, as we have managed to cover >80% of the world’s population already.
• Get a better sense of the magnitude of the antifungal resistance problem and the resulting health burden by, e.g., literature review and talking to experts. We don’t have a good sense of the current health burden of antifungal resistance and we are not sure whether anyone has tried to estimate this.

Tractability/interventions:

• Get a better sense of the tractability of specific interventions (e.g., through literature review and talking to experts), whether they are suited for philanthropic funding/grantmaking, how they could be supported more concretely, and whether they are scalable.
• Understand more about the impact of BARDA adding fungal medical countermeasures to their areas of interest, and the WHO priority pathogens list and actions planned as a result.
• Identify the top three most promising interventions and do a slightly deeper dive into what an intervention could look like and what the impacts might be. For example:
  • If we focused on antifungal vaccines, we could identify better what current efforts are, how much it would cost to develop a vaccine and how many years this could take. Moreover, we could look for estimates or do a back-of-the-envelope calculation on potential lives/DALYs saved due to the vaccine.
  • If we focused on antifungal stewardship, we could identify, e.g., whether there are any important technical or political barriers, whether there are existing lobbying efforts that could be scaled, and whether there are important downsides.
    • Example questions we might consider include: Is a reduced use of azoles in agriculture possible/feasible? Could it have unintended side effects on productivity/food security?
• Add one to two short case studies on promising current interventions, e.g.:
  • Case study on GAFFI’s diagnosis program demonstration in Guatemala, which increased screening of HIV patients for fungal diseases and reportedly reduced mortality by 7 percentage points.
  • Case study on some potentially promising R&D efforts:
    • One option for this could be a pan-fungal vaccine candidate, which has been promising in preclinical animal models of invasive aspergillosis, invasive candidiasis, and pneumocystis pneumonia (mentioned here).

Acknowledgments

Jenny Kudymowa and James Hu jointly researched and wrote this report. Jenny also served as the project lead. Tom Hird supervised the report. Special thanks to Melanie Basnak, Aisling Leow, and Sam Donald (Open Philanthropy) for helpful comments on drafts. Thanks also to Adam Papineau for copy-editing and Rachel Norman for assistance with publishing the report online.

Further thanks to David Denning (Global Action for Fungal Infections [GAFFI]), Juan Luis Rodríguez Tudela (GAFFI), Tom Chiller (Centers for Disease Control and Prevention), Marcio Rodrigues (Fiocruz), and Arturo Casadevall (Johns Hopkins Bloomberg School for Public Health) for taking the time to speak with us. 

Open Philanthropy provided funding for this report, but it does not necessarily endorse our conclusions.

We invite you to explore more RP research via our database and stay updated on new work by subscribing to our newsletter.

Appendices

Appendix A: Fungal diseases 101

Certain species in the fungal kingdom can cause diseases — known as fungal diseases, fungal infections, or mycoses — in animals, including humans. While fungi are ubiquitous in type and quantity, less than 0.5% (<500) of fungal species have been known to cause disease in humans (Richardson & Warnock, 2012, p. 5). Less than 0.1% (<100 species) are currently known to cause disease in people without underlying illness.

Fungi rank among the oldest recognized causes of disease, with records documenting possible cases since the second millennium BCE (Ainsworth, 2002, pp. 1-3; Homei & Worboys, 2013, p. 6). The discipline of medical mycology, however, was only established in the 20th century, gaining more attention with the advent of immunosuppressive therapies in the 1950s and 1960s and the HIV/AIDS epidemic in the 1980s (Homei & Worboys, 2013, pp. 9-11). It is widely considered to be a neglected field (“Stop neglecting fungi”, 2017; Homei & Worboys, 2013, p. 1; Rodrigues, 2020; Williams, 2022).

How are fungal diseases categorized?

Fungal diseases can be categorized according to several schemata.⁴² One schema, used by Richardson and Warnock’s (2012) academic textbook, considers the initial site of infection (pp. 5-7). Mycoses can be “superficial” (infection of the outermost layers of skin, nails, and hair and of the mucous membranes), “subcutaneous” (that of deeper layers of skin, tissue, and bone), or “systemic” (that of the internal organs and blood).⁴³

Certain fungal diseases (and pathogens) have received special designations pertaining to their neglectedness or priority. For example, the WHO’s list of neglected tropical diseases (NTDs) includes mycetoma and chromoblastomycosis (WHO, n.d.), while GAFFI further advocates for the inclusion of sporotrichosis, paracoccidioidomycosis, and fungal keratitis among NTDs (GAFFI, n.d.). The WHO (2022) and the US CDC (2019) have also released lists of priority pathogens for public health and antifungal resistance significance.

What causes fungal diseases, and who is most at risk from them?

Fungal diseases are generally caused by the inhalation, ingestion, or traumatic implantation of fungi that grow in the environment (Richardson & Warnock, 2012, p. 5). Inhaling the spores of a fungal pathogen generally does not cause serious infection in immunocompetent people,⁴⁴ but traumatic subcutaneous implantation can cause serious infection and lifelong disability even in the immunocompetent.⁴⁵ Other fungi are commensal, or hosted on healthy human skin and mucous surfaces, and only cause disease when the host has an “imbalance.”⁴⁶

Some fungal diseases are endemic to certain regions of the world, as determined by the habitat of the relevant fungal pathogens.⁴⁷ While some of the most common species are found worldwide, mycetoma is, for example, endemic to the so-called “Mycetoma Belt” near the equator (WHO, 2022). The increasing incidence of certain fungal infections has been linked to international travel (Richardson & Warnock, 2012, pp. 8-9; Ericsson et al., 2002).

Most life-threatening, invasive, or systemic fungal diseases occur in people with diseases that cause immunodeficiency — such as HIV/AIDS and diabetes — and those receiving immunosuppressive care — including cancer chemotherapy and solid organ transplants (Richardson & Warnock, 2012, pp. 6-8). Non-life-threatening fungal diseases, including ringworm and thrush, are extremely common and can affect those who are generally immunocompetent. However, people with weakened immune systems are still at greater risk of infection upon contact (CDC, n.d.).

How are fungal diseases diagnosed?

Fungal diseases are diagnosed through a combination of clinical observation and laboratory testing (Richardson & Warnock, 2012, pp. 9-13). Superficial fungal diseases often present with distinguishing lesions. However, deeper infections usually have non-specific clinical and radiographic presentation and affect those with underlying conditions (Lass-Flörl, 2017, p. 3).⁴⁸ Laboratory-based methods that involve identifying the species of fungus are required to reach a definitive diagnosis (Lackner & Lass-Flörl, 2013).

There are three types of laboratory method: conventional, serological, and molecular (see Tables 1 and 2 in Lass-Flörl, 2017, pp. 4-5). Conventional methods include microscopic examination (including histopathology) and culture.⁴⁹ Serological methods involve the detection of fungal cell wall components, antigens, and antibodies (but not serum antibodies) in blood and other body fluids.⁵⁰ Molecular methods involve testing for fungal genetic information (generally DNA) in clinical specimens.⁵¹ However, high costs and a lack of skilled personnel mean that the use of serological and molecular methods is far less widespread in LMICs than in HICs (Arastehfar et al., 2019).

How are fungal diseases treated?

Fungal diseases that do not self-resolve are treated with antifungal drugs. There are four⁵² main classes of antifungal in systemic, clinical use: polyenes, (tri)azoles, echinocandins, and the pyrimidine analogue flucytosine (Fisher et al., 2022; Houšť et al., 2020). GAFFI has developed a priority list of six antifungals (GAFFI, n.d.), while the most recent WHO Essential Medicines List includes eight (WHO, 2021, p. 17).⁵³ Kneale et al. (2016) found that antifungal drugs varied drastically in price and availability by country.⁵⁴

Our understanding is that the appropriate treatment is largely determined by the fungal pathogen, but still varies somewhat across diseases caused by the same pathogen. To take the various Aspergillus infections as an example: chronic pulmonary aspergillosis’s first-line treatments are itraconazole or voriconazole (Maghrabi & Denning, 2017); invasive aspergillosis’s are voriconazole and isavuconazole (Jenks & Hoenigl, 2018); and allergic bronchopulmonary aspergillosis’s is glucocorticoids, with itraconazole also recommended in some cases (Jack & Bajaj, 2022).

Any given antifungal is efficacious for several fungal diseases, albeit to varying degrees (see GAFFI, 2015, p. 11, for a likely slightly outdated overview of the efficacy of different antifungals by disease). For example, in 2015, six different antifungals had some efficacy against (different types of) aspergillosis (ibid, p. 11). Half of those antifungals had low efficacy, and the other half had moderate efficacy (with no antifungal having maximum efficacy).

What is antifungal resistance?

Antifungal resistance⁵⁵ denotes the non-susceptibility of fungal pathogens to drug treatment, which is a growing problem linked to the overuse of antifungal drugs in both clinical and environmental contexts (Fisher et al., 2022, p. 557). Resistance emerges in an evolutionary, mechanistic process whereby cells undergo genetic changes that affect drug-target interactions and can be acquired in vivo — i.e., during therapy — or in the environment, mostly in agriculture (p. 558).⁵⁶ High-temperature ecological hotspots such as composts, green waste stockpiles, and greenhouses can amplify environmental resistance (p. 563).

In a loose parallel to fungal disease diagnosis, resistance can be identified through susceptibility testing with culture and, increasingly, with molecular tools. Microdilution reference methods with cultured specimens are considered the “gold standard” but are labor- and time-intensive and, thus, infrequently performed; moreover, Europe and the US use different clinical “break points” to measure susceptibility, thus compromising the process’s standardized use (Fisher et al., 2022, pp. 558-563). Molecular tools including PCR assays are advancing but still under development (pp. 563-564).

In response to rising antifungal resistance, antimicrobial resistance surveillance initiatives such as JPMIAR and WHO GLASS have begun to include fungal pathogens, although most surveillance programs still exclude fungi (Fisher et al., 2022, pp. 564-565).

Appendix B: Summary of our death and DALY estimation process

Stage 1: Quick estimation of the death and DALY burden for 30 fungal diseases

Incidence:⁵⁷

• We used the most recent incidence estimates we found (FIT, 2020) and multiplied them by a “subjective underdiagnosis inflator” for each disease to account for missed cases due to underdiagnosis. We assumed 1.1x as a lower bound for the inflator, as even in the absence of underdiagnosis, we expect that cases have increased due to population growth (as many fungal disease burden estimates are ~10 years old). As an upper bound, we used an expert’s stated inflation factors where available, and assumed 2x otherwise. For example, we increase the invasive aspergillosis incidence by a factor of 1.1-3x (80% CI).

Deaths:

• We calculated the number of deaths per disease in different ways depending on data availability.
  • We calculated deaths per year by either:
    • multiplying incidence cases and mortality rates, which are based on FIT (2020) and the LIFE website factsheets (LIFE, 2022).
    • taking death estimates from GAFFI (2016) and applying our underdiagnosis inflation factor.
  • According to Juan Luis Rodríguez Tudela (a board trustee at GAFFI), mortality rates (at least for some diseases) have decreased in recent years. Therefore, we reduced our upper-bound mortality rates by one-third.⁵⁸
  • Where we were not able to find incidence or death estimates for some (probably smaller burden) fungal diseases, we assumed 1,000-15,000 deaths each (80% CI). This is merely a guess.

Disability weights:

• We chose disability weights from a catalog of similar diseases (in terms of symptoms) in the 2019 GBD study (IHME, 2020b). We did not have time to do this for all diseases (especially the ones with a likely smaller burden), so for some diseases we just assumed a disability weight of 0.05. This is just a guess.

Years of life lost (YLLs):

• We originally devised priors for YLLs (7-25 years [80% CI]) as described in this footnote.⁵⁹
• However, we later discarded those priors and updated our calculation of YLLs for all fungal diseases based on further research, as described in Stage 3.

Years lived with disability (YLDs):

• For diseases for which prevalence figures are available, we estimated YLDs by multiplying disability weights by the prevalence of a disease.
• For other diseases, we multiplied disability weights by the incidence of a disease and assumed an average duration of illness of 0.5-3 years (80% CI). This is merely a guess and we expect there to be substantial heterogeneity across diseases.

Stage 2: Refining our death estimates for chronic pulmonary aspergillosis and invasive candidiasis/candidemia

Chronic pulmonary aspergillosis (CPA):

We decided to leave our initial estimates of CPA prevalence and mortality rates from Stage 1 unchanged. While we have a lot of uncertainty around those estimates, it is not clear to us whether the estimates are over- or underestimates. Thus, any adjustment of those figures would seem arbitrary to us.

• A “conservative” estimate of deaths per year that we’ve seen reported in literature is 450k (see Brown et al., 2012), which is based on a mortality rate of 15%, applied to a prevalence of 3M. We tried to re-evaluate both components:
• Prevalence:
  • We are skeptical about the prevalence estimate of 3M, for three reasons: (1) We have not been able to trace back the original source of this estimate, as the chain of citations we tried to follow didn’t lead anywhere. We think it might be a rough extrapolation mainly based on data from CPA following tuberculosis (1.2M cases; Denning et al., 2011), but we weren’t able to evaluate its validity. (2) Tom Chiller suggested these figures seemed too high. (3) We have not seen any other global prevalence estimates for CPA.
  • However, some recent country-level prevalence estimates are much higher than previous estimates. For example, Denning et al. (2022) estimated CPA five-year prevalence in India at ~1.5M in TB patients (compared to a previous estimate of ~290k). We have not seen the more recent national prevalence estimates aggregated to a global estimate. We reviewed Denning et al. (2022) only superficially, but it was published in a peer-reviewed journal and seems of high quality to us, at least at first glance.
  • This leaves us highly uncertain about the prevalence. In the absence of other global prevalence estimates, we don’t have any other estimates to rely on, and we did not have sufficient time to re-estimate the global prevalence based on national studies (nor did we have time to evaluate the quality of the national studies).
• Mortality rate:
  • We found several reasons to believe that mortality rates might be either overestimated or underestimated and we did not find either side of arguments more convincing than the other.
  • Mortality rates might be overestimated due to:
    • Selective reporting of hospitalized cases:
      • There is some concern among experts that mortality rates might be inflated by “selective reporting of hospitalized cases.”⁶⁰ Unfortunately, we found relatively little evidence to support or refute this concern. We have not been able to find data on the shares of hospitalized vs. nonhospitalized CPA patients.⁶¹ The best recent review of CPA mortality rates (Denning et al., 2022, Fig. 2) shows that the majority of mortality studies are based on outpatient cohorts, which alleviates our concern that mortality rates are mainly estimated in inpatient settings in very ill populations.
    • Representing “overall” rather than “attributable” mortality:
      • The 15% mortality rate figure comes (we think) from Denning et al. (2011), and it seems to be more of an assumption rather than an estimate, though we aren’t quite sure. We reviewed more recent evidence:
        • The best recent reviews of mortality of CPA patients we’ve seen are Lowes et al. (2017) and Denning et al. (2022), who report the survival of CPA patients from several studies,⁶² all from HICs.⁶³ One-year mortality ranges from ~10-40%, and five-year mortality from ~15-85%. However, both of those reviews show overall mortality rates, rather than CPA-attributable mortality. According to our discussion with Rodríguez Tudela and Denning “the majority of these deaths are attributable to CPA,” but they did not explain how they came to this conclusion.
        • Very few studies estimate attributable mortality, e.g., Nam et al. (2010) find a five-year CPA-attributable mortality of 42% (51% overall mortality). Furuuchi et al. (2018) find a CPA-attributable mortality of 27% at seven years (75% overall mortality). Neither of those studies explains how the attribution was made.
      • Thus, if overall one-year mortality is on average 25% (based on Lowes et al., 2017, and Denning et al., 2022), then a 15% attributable mortality seems plausible to us.
  • Mortality rates might be underestimated due to:
    • Being predominantly based on data from high-income countries:
      • See Denning et al. (2022) and Lowes et al. (2017).
    • Underdiagnosis of mortality:
      • According to Rodríguez Tudela, it is wrong to assume that known fungal disease deaths are near to true fungal disease deaths due to misdiagnosis of cases discovered at autopsy.⁶⁴

Invasive candidiasis and candidemia (IC/C):

Following further research, we elected to replace our Stage 1 IC/C estimate. We use a “conservative estimate” of incidence of 650k, based on our own calculations using recent incidence studies from various countries, which are detailed further in Appendix D. We use mortality rate estimates of 10-50% (80% CI) based on several studies on attributable mortality for IC/C.

Incidence:

• As there is a large variation in national incidence rates of IC/C (and no estimates are available for many countries), global incidence could be, according to Rodríguez Tudela and Denning, anywhere between 400k and 4.8M cases.⁶⁵ As some of the populous countries, like India and Brazil, have high incidence rates (FIT, 2020), the true incidence could be higher than 900k. These figures sound plausible to us, if we trust national incidence estimates (which we have not reviewed). Our own very rough calculations suggest that it could be ~650k,⁶⁶ which we decided to use as our ‘conservative estimate.’

Mortality rate:

• The most commonly cited mortality rate estimates (30-65%) are based on a LIFE factsheet and Brown et al. (2012). At closer inspection, we found these mortality figures to be rather opaque and we’re not completely sure where they originate from. We expect that these figures refer to overall rather than attributable mortality.
• Thus, we reviewed the literature on attributable mortality:
  • Zaoutis et al. (2015) estimated attributable mortality to candidemia in the US to be ~10-15% (for children and adults, respectively), based on a retrospective cohort study and propensity score matching. Tom Chiller also pointed us to attributable mortality rates of ~10-20% (mentioning the same study). We expect that mortality in LMICs is higher than that. Another study that supports these figures is Pappas et al. (2018) who state that the “attributable mortality among all patients with candidemia has been reported to be between 10% and 47%, but a more accurate estimate is probably 10-20%.”
  • We found several other studies that report attributable mortality rates to IC/C, e.g.:
    • Gudlaugsson et al. (2003): attributable mortality: 49%; crude mortality: 61%. This study refers to “nosocomial candidemia.”
    • Chakrabarti et al. (2014): attributable mortality: 20%; crude mortality: 45%. This study focuses on ICU-acquired candidemia in India.
    • Falagas et al. (2006): attributable mortality: 5%-71%. This is a systematic review study; all underlying studies are based on hospitalized patients in ICUs or hospital wards.
• Based on these figures from the literature, we decided to use mortality rate estimates of 10-50% (80% CI).

Stage 3: Refining our YLL estimates for chronic pulmonary aspergillosis and invasive candidiasis/candidemia

The following steps describe how we estimated new counterfactual life expectancies for CPA and IC/C, in order to adjust our YLL/DALY estimates for them:

1. Identify the main underlying conditions for CPA and IC/C.
2. Assign a proportion of CPA and IC/C patients to each underlying condition, and estimate average survival time following diagnosis of each underlying condition.
3. Calculate a weighted average of years lived after diagnosis of underlying conditions, then halve the calculated year range (assuming CPA incidence occurs at the midpoint of underlying disease progression) to yield counterfactual life expectancy (LE) for each fungal disease.

Step a

Table D1 summarizes the main risk factors for CPA and IC/C. We find that CPA is mainly associated with TB, COPD/emphysema, and possibly lung cancer, and that IC/C is mainly associated with immune deficits and critical care. However, the evidence base for this is poor and quantitative estimates are low quality.

Table D1: Proportion of patients with underlying conditions by fungal disease

	Chronic pulmonary aspergillosis67	Invasive candidiasis/ candidemia68
HIV/AIDS	Unknown; likely only a small minority69	Unknown; likely only a small minority70
Previous cancer and cancer treatment	No formal estimate; previous lung cancer — possibly observed at a frequency of up to ~5% — elevates risk71	Unknown; haematologic malignant disease, solid-organ transplantation, and solid-organ tumors elevate risk72
TB	~17%-80%; risk is particularly elevated for drug-resistant TB73	Unknown; it is possible TB could elevate risk74
Other	“Vast majority” have pre-existing chronic lung conditions, including, e.g., ~30%-50% COPD/emphysema75	One estimate puts the proportion with immune deficits at ~60% and those in critical care at ~30%; risk appears particularly elevated for those experiencing long-term ICU stays76
None	Negligible (by implication)	~5%77

Step b

For CPA, we focus on TB, emphysema/COPD, and lung cancer.

• We assign TB 53% weight.⁷⁸ We estimate LE upon diagnosis of TB is 5-15 years. Without treatment, LE is less than 10 years;⁷⁹ with treatment, LE appears normal, although there is evidence to suggest there is a longevity penalty of ~3.6 years.⁸⁰ However, a Nature Reviews primer (Pai et al., 2016) indicates that the cure rate of TB is only ~50% in ideal situations and that of drug-resistant TB is only ~20% in low-resource settings. The low cure rate in low-resource settings suggests a closer counterfactual LE for most CPA patients (99% in LMICs; see Figure E1) to the non-treatment case.
• We assign COPD 42% weight. We estimate LE upon diagnosis of COPD is 3.4-9.7 years. Surprisingly, we were not able to find a good estimate of COPD patients’ survival rates in the academic literature after 30 minutes of research. A Dutch study suggests that among patients admitted to a hospital for “COPD exacerbation,” LE is 3.4 to 9.7 years, depending on disease stage (Van Hirtum et al., 2018).
• We assign lung cancer 5% weight. We estimate LE upon diagnosis is 2-5 years. Even in an HIC like the UK, lung cancer has a very poor prognosis, with most patients not surviving more than one year after diagnosis (Cancer Research UK, 2022). However, it is possible that the group of lung cancer patients who are separately diagnosed with CPA have survived with lung cancer for longer and constitute a separate prognostic category.

  For IC/candidemia, we focus on critical care and immune deficits.

  • We assign critical care 51% weight. We estimate LE upon admission is 2-5 years. Critical care again covers a very wide range of scenarios, from neonatal care and care for otherwise healthy people who have suffered traumatic injury — i.e., two groups who likely have longer counterfactual life expectancies — to care for those with terminal illnesses. We infer, based on an age profile of ICU patients in a Canadian province showing that old patients are much more likely to be admitted (Garland et al., 2013, p. 4), that the latter group dominates. This dominance by very elderly patients implies a very low (very likely <5 years) average counterfactual LE among this subgroup of IC/candidemia patients. We make the (strong and uncertain) assumption that globally, the demographics of critical care patients will be similar to than in Canada.
  • We assign non-HIV/AIDS immune deficits 43% weight. We estimate LE upon diagnosis is 4.9-17.5 years. Immune deficiencies are a very broad medical category covering autoimmune disorders, neutropenia, and other diseases; it is thus extremely difficult to estimate a counterfactual LE for this category. In the absence of information about the specific immune deficits, the 7-25 year interval, devised as a blanket input in our original model, seems somewhat too high. This is because our impression is that, in the era of antiretroviral therapy, most immune deficits are harder to manage than HIV/AIDS. We therefore apply a 30% discount to the interval of 7-25 years to obtain 4.9-17.5 years.
  • We assign “no underlying conditions” 6% weight. We somewhat arbitrarily assume a counterfactual LE of 10-25 years.

Step c

On the basis of Steps a and b, we adjust our original estimate by calculating a weighted average of the LE-upon-diagnosis intervals for each disease’s main underlying conditions. Our calculations are laid out here (CPA) and here (IC/C).

• For CPA, we calculate the weighted average of 5-15 years (53%), 3.4-9.7 years (42%), and 2-5 years (5%). This yields a weighted average LE upon diagnosis for CPA underlying conditions of 4-12 years; assuming CPA diagnosis occurs at the midpoint of disease progression, we obtain a counterfactual LE for CPA of 2.1-6.1 years (80% CI).
• For IC/C, the calculation is somewhat different given the inclusion of a “no underlying conditions” category. We therefore calculate a weighted average of 2-5 years (51%), halved; 4.9-17.5 years (43%), halved; and 10-25 years (6%), not halved. This yields an estimated counterfactual LE for IC/C of 2.2-6.5 years (80% CI).
• We did not have time to repeat this exercise for all other fungal diseases, so for every other fungal disease — except skin, hair, and nail fungal infections — we took the average of the values for CPA and IC/C (2.1-6.3 years, [80% CI]). However, we are highly uncertain about this.
  • We took this decision because we very briefly (~15 minutes search) reviewed the main underlying conditions for some of the other fungal diseases with a relatively high mortality (e.g., histoplasmosis) and found these to be similar to CPA and IC/C. We expect that this indicates a similar counterfactual LE.

Appendix C: Main populations affected and size of antifungal resistance problem

Table C1: Main populations affected and size of antifungal resistance problem by fungal disease

Disease category	Main risk factors & populations affected	Main geographical regions affected	Main age groups affected	How big is the antifungal resistance problem?
Chronic pulmonary aspergillosis (CPA)	Respiratory diseases (e.g., tuberculosis, asthma, lung cancer) (LIFE - CPA factsheet)	Highest prevalence rates in sub-Saharan Africa, Asia, and Greenland (GAFFI Burden of disease maps)	N/A81	Preferred treatment: Itraconazole and voriconazole. Antifungal resistance: “Azole resistance in A. fumigatus is becoming an increasing problem [...]” (LIFE - CPA factsheet)
Progressive disseminated histoplasmosis (PDH)	Immunosuppression, HIV/AIDS (LIFE - PDH factsheet)	The Histoplasma fungus is present worldwide but is most common in North America and Central America (CDC Histoplasmosis maps)	Extremes of age (LIFE - PDH factsheet).	Antifungal resistance: N/A
Candidemia and invasive candidiasis	Immune deficiencies (e.g., due to prematurity in infants, chemotherapy, pancreatitis, burns, multiple antibiotics ) & patients with longer stays in intensive care, such as COVID-19 patients (LIFE - Candidemia factsheet)	~50% of candidemia cases in Asia. Highest incidence in LMICs , particularly Brazil, Pakistan, Qatar, and Thailand (Bongomin et al., 2017, p. 5).		Preferred treatment: Echinocandins. Sometimes fluconazole or amphotericin B. Antifungal resistance: “Antifungal resistance is especially a problem with fluconazole. Resistance to echinocandins is not frequent but case reports and small series have been described” (LIFE - Candidemia factsheet).
Pneumocystis pneumonia	HIV/AIDS, transplants, leukaemia, lymphoma, malnourished children (LIFE - PCP factsheet)	N/A	N/A	N/A
Invasive aspergillosis	Acute leukaemia, organ transplants, respiratory diseases or autoimmune disorders, or in intensive care (LIFE - IA factsheet)	N/A	N/A	Preferred treatment: Voriconazole. Antifungal resistance: N/A
Severe asthma with fungal sensitization (SAFS)	Asthma (LIFE - SAFS factsheet)	“Probably worldwide” (LIFE - SAFS factsheet)	N/A	Preferred treatment: Itraconazole. Antifungal resistance: N/A
Cryptococcal meningitis (CM)	Immunocompromised patients (e.g., due to AIDS, transplants) (LIFE - CM factsheet)	“More common in sub-Saharan Africa and tropical countries (Brazil, Thailand, Malaysia, Papua New Guinea)” (LIFE - CM factsheet).	N/A	Preferred treatment: Amphotericin B + flucytosine. Low level of resistance for flucytosine in Cryptococcus (Denning, 2021).
Mucormycosis	Patients with, e.g., immunosuppression, organ transplants, diabetes, malnutrition and prematurity, IV drug abuse (LIFE - Mucormycosis factsheet)	“Mucormycosis is much more common in India than most other countries” (LIFE - Mucormycosis factsheet).	N/A	N/A
Skin, hair, and nail fungal infections (SHN)	“The vast majority of people with fungal infection of skin, hair and nails are otherwise healthy, but a small group are immuno- compromised” (LIFE - SHN factsheet)	N/A	N/A	N/A
Vulvovaginal thrush	Thrush is very common. Risk factors are pregnancy, antibiotic use, diabetes mellitus, cystic fibrosis (LIFE - Vulvovaginal thrush factsheet)	Prevalence highest in high- and middle-income countries (GAFFI Burden of disease maps)	“By a mean age of 24 years, 60% of women had suffered at least one episode of vulvovaginal candidosis; 36% had at least one episode a year; 3% had it ‘almost all the time’” (LIFE - Vulvovaginal thrush factsheet).	Preferred treatment: Azoles. Antifungal resistance: “Recurrent use of azole antifungal agents can lead to colonization with azole-resistant Candida” (LIFE - Vulvovaginal thrush factsheet).
Candida peritonitis	“Follows abdominal surgery, pancreatitis or other intra-abdominal sepsis, or as a complication of peritoneal dialysis. [...] Other risk factors include multiple antibiotics and diabetes mellitus” (LIFE - CP factsheet)	N/A	N/A	Preferred treatment: Echinocandins or fluconazole. Antifungal resistance: N/A

Appendix D: Fungal disease burden (% of total cases) by region

Methodological note

The best source for country-level fungal disease case numbers we could find was Bongomin et al. (2017). However, its data-analytic methods are likely flawed.⁸² Moreover, an email from the corresponding author suggested that we de-emphasize those numbers in our analysis. We therefore disregard Bongomin et al.’s (2017) charts and focus on the raw incidence data presented in tables.

We supplemented⁸³ and reanalysed this data (see calculations here) to estimate the World Bank income classification (Figure E1) and continental (Figure E2) distribution of the total burden of five of the top fungal diseases. Our analysis ultimately includes data from 64 countries, though we do not have data on all five fungal diseases in every country.

The individual studies (both those in Bongomin et al., 2017, and the additional studies we used) only provide rough, back-of-the-envelope-type estimates. In addition, they could be highly skewed by differences in diagnostic efforts. We therefore have low confidence in the results of the analysis. Moreover, not all estimates are from the same year, though we think this is unlikely to meaningfully affect the results. Also note that raw case numbers used in our calculations for this section are not adjusted for the rate of underdiagnosis and consequently do not reflect our view of the distribution of the burden among diseases.

One such study is Ray et al. (2022), which estimates the burden of fungal diseases in India by layering multiple assumptions about the rates of diseases in people who have other conditions and drawing from estimates of those conditions. For example, this is one of several calculations that feed into an estimate for invasive aspergillosis incidence:

Total cases of acute myeloid leukemia (AML) were estimated to be 40% of the total burden of leukemia and myeloma. The annual incidence of acute myeloid leukemia is 25 224 cases [39]. The rate of IA [invasive aspergillosis] in AML patients receiving induction chemotherapy was 13% [27], and an equal number of cases of IA was assumed in all other cases of leukemia and myeloma [28]. (Ray et al., 2022, p. 2)

Another individual study is Zhou et al. (2022), for China. This paragraph describes its estimate of chronic pulmonary aspergillosis prevalence:

We used the WHO 2017 figures for PTB [pulmonary tuberculosis] to calculate CPA [chronic pulmonary aspergillosis] (11). We calculated CPA incidence after PTB based on our previous estimate, assuming that 22% of patients are left with a pulmonary cavity and that 22% of these patients develop CPA each year, as did 2% of those without a cavity (30). This calculation derives an annual incidence of CPA, which we converted to a 5-year period prevalence by assuming a 15% annual death or surgical resection rate. Given that PTB is one of several underlying causes of CPA, we conservatively assumed that PTB was primarily responsible for 33% of all CPA cases (31). (Zhou et al., 2020, p. 2139)

Results

Most of the major fungal diseases appear to have higher caseloads in Asia due to a combination of its large population size (~59% of the world) and higher cases per 100k inhabitants.

• Asia appears to have significantly higher cases per 100k of chronic pulmonary aspergillosis and candidemia⁸⁴ than other continents. Consequently, Asia comprises ~78% and ~66% of total cases, respectively.
• Asia appears to have an order-of-magnitude higher cases per 100k of invasive aspergillosis than other continents. Consequently, Asia comprises ~93% of total cases.
• A notable exception seems to be Pneumocystis pneumonia, of which Africa appears to have an order-of-magnitude higher cases per 100k than other continents. Consequently, Africa comprises ~61% of total cases.

Most of the major fungal diseases appear to have higher case loads in LMICs due to a combination of their large collective population (~84% of the world) and higher cases per 100k inhabitants.

• Chronic pulmonary aspergillosis, candidemia, invasive aspergillosis, and Pneumocystis pneumonia cases are ~99%, ~88%, ~97%, and ~95% in LMICs.
• However, only ~74% of severe asthma with fungal sensitization cases are in LMICs, due to higher prevalence of asthma in HICs.

Progressive disseminated histoplasmosis could disproportionately affect the Americas (where it is hyperendemic) and Africa — and by extension LMICs (see Figure E3). However, data is largely unavailable at the country level and so is excluded from the foregoing analysis.

Figure E1: Burden share (% of total prevalence or incidence) of top fungal diseases by World Bank income classification

Note. Reanalysis of data from Bongomin et al. (2017) and newer studies collated by GAFFI. Full data and working in this spreadsheet.

Figure E2: Burden share (% of total prevalence or incidence) of top fungal diseases by continent

Note. Reanalysis of data from Bongomin et al. (2017) and newer studies collated by GAFFI. Full data and working in this spreadsheet.

Figure E3: Estimated areas with histoplasmosis

Note. From Ashraf et al. (2020).

Appendix E: Explanation of potential future incidence growth of fungal diseases

From talking to experts, we learned that a very rough lower-bound estimate of the future burden of fungal diseases could be made by extrapolating past trends of major risk factors (e.g., cancer, HIV/AIDS, organ transplants, aging, respiratory diseases). Thus, we considered past trends of several important risk factors as a proxy for future trends and found that:

• The combined incidence of cancers, HIV/AIDS, and tuberculosis, chronic respiratory diseases, and diabetes mellitus grew annually by 0.8% in the 10-year period until 2019 (GBD Results Tool, 2019).
• The annual growth rate of global organ transplants (per 1M inhabitants) from 2010-2021 was 3.1% (GODT data, 2021).
• The global population share aged 65+ grew annually by 2% from 2011-2021 (World Bank Data, 2021).

Unfortunately, we don’t have a good sense of how exactly an increase in these risk factors would translate into an increase in fungal disease cases. Our general understanding of the literature is that, while these are known risk factors, their exact relationship with fungal diseases is less well understood, and fungal diseases can also occur in healthy individuals. Thus, we don’t expect that, for example, a growth in risk factors will necessarily lead to a growth in fungal disease incidence at the same rate. Therefore, we (somewhat arbitrarily) assume that due to an increase in risk factors, the incidence rate of fungal diseases will increase by 0.2%-1.2% per year (80% CI).

Note that these figures only refer to growth in incidence. We expect that there will be a countervailing effect on fungal disease incidence/mortality due to likely improvements in treatment of risk factors. We did not have sufficient time to consider the likely extent of treatment improvements in the future and how this would impact fungal disease incidence and mortality.

Appendix F: Selected research groups focused on fungal diseases

• Mycotic Diseases Branch (Centers for Disease Control and Prevention)
• MRC Centre for Medical Mycology (University of Exeter)
• AReST [Antifungal Resistance: From Surveillance to Treatment] (Leibniz Institute for Natural Product Research and Infection Biology, Hans Knöll Institute)
• AFRICA CMM Medical Mycology Unit⁸⁵
• IMPRINT [International Mycoses Prevention, Research, Implementation, Networks and Training] (London School of Hygiene and Tropical Medicine)
• Manchester Fungal Infection Group (University of Manchester)
• Fungiscope (University of Cologne)
• Fungal Kingdom: Threats & Opportunities Research Program (CIFAR)

Appendix G: Calculation of cost-effectiveness of providing treatment for CPA and IC/C

See here for our calculations.

Chronic pulmonary aspergillosis:

• Treatment:
  • First-line treatment: Oral itraconazole 400 mg daily
  • Duration: Minimum is six months’ treatment. There is evidence that 12 months might be better (Warris & Armstrong-James, 2022; Maghrabi & Denning, 2017; Sehgal et al., 2022)
• Treatment effectiveness:
  • The CPA treatment effectiveness literature can be broadly grouped into three types of measures: (1) response rates, (2) mortality rates, and (3) quality of life measures:⁸⁶
    • Response rates: The most convincing figures on treatment effectiveness we found are on response rates (%) to treatment, and are reported to be ~30%-93% (usually 60%-70%) (Izumikawa, 2012, slide 13; Agarwal et al., 2012, p. 565; Felton et al., 2022; Felton et al., 2010). This range was confirmed by experts we spoke with.⁸⁷
    • Mortality rates: We reviewed 4 of the 20 studies related to mortality effects of CPA treatment based on input from Denning and Rodríguez Tudela (Denning et al., 2022; Nam et al., 2010; Jewkes et al., 1983; Tomlinson & Sahn, 1987),⁸⁸ but our understanding is that none of these have a control group which would allow us to know how treatment affects mortality:
      • However, an important finding in Nam et al. (2010) is the “poor prognosis of these patients despite antifungal treatment, mainly oral itraconazole. Itraconazole therapy was continued for more than 3 months in only 60% (26/43) of the patients, because of the high early mortality rate. Moreover, in these 26 patients, clinical or radiological improvement was observed in only about 60%. Fifty percent of all patients died, and the median survival time was only 62 months. [...] Itraconazole may be a suppressive, rather than curative, therapy in CNPA.”
  • Overall, our impression is that symptoms improve at least to some extent in 60%-70% of cases, but mortality effects seem to be pretty unclear/limited. Treatment rarely leads to a cure, and often needs to be taken for years, as relapse is common. For the sake of providing a best guess, we assume that half of the one-year mortality (estimated at ~15%), can be prevented with treatment, such that 0.08 lives are saved per treatment⁸⁹
• Cost of treatment (see calculations here):
  • We obtained data on the minimum and maximum cost per daily dose of treatment from GAFFI’s database on antifungal prices for itraconazole. We collected this information for the 10 LMICs with the highest number of estimated CPA cases (see column P here).
  • Treatment cost per case:
    • Assuming a treatment duration of 12 months and using the cheapest drug price per country, the treatment cost per case ranges from $47 in Vietnam to $59k in China.
  • Treatment cost per life saved:
    • Assuming a treatment duration of 12 months, using the cheapest drug price per country, and assuming that there are 0.08 lives saved per treatment, the treatment cost per life saved ranges from ~$630 in Vietnam to ~$780k in China.
• Barriers to treatment:
  • High side effects: “Itraconazole is not free from side effects and can cause considerable toxicity reportedly in 40–50% of patients. Adverse effects include gastrointestinal upset, hair loss, peripheral neuropathy, hypertension and ankle oedema which may or may not be an early sign of congestive heart failure” (Maghrabi & Denning, 2017).
    • “However, these side effects may be lower in LMICs since the patients are much younger and are more likely to tolerate these drugs” (Bongomin et al., 2016).
  • “All patients with CPA require long-term surveillance because relapse is common and cure is rare” (The Lancet Respiratory Medicine, 2016).
  • Therapy with itraconazole requires “therapeutic drug level monitoring, which is impractical in most centres in sub-Saharan Africa” (Bongomin et al., 2020).
  • According to Bongomin et al. (2020), “diagnosis of CPA is challenging and requires a systematic approach to assessment and interpretation of findings, both of which are necessary for correct disease classification and selection of targeted antifungal treatment and duration of said treatment. Diagnosis of CPA requires a corroboration of a combination of diagnostic armamentaria i.e., Clinical, Radiological, Immunological and Mycological modalities (CRIM).”

Invasive candidiasis/candidemia:

• Treatment:
  • First-line treatment: Echinocandins (caspofungin, micafungin, anidulafungin) given through the vein; “the choice between the three echinocandins in real life is usually based on cost” (Tagliaferri & Menichetti, 2015).
  • Duration: At least 14 days after the last positive blood culture and sometimes even longer (Tagliaferri & Menichetti, 2015); Neoh et al. (2013) assume 14-60 days.
• Treatment effectiveness:
  • A US population-based study (Pfaller & Diekema, 2007) found an attributable mortality rate of 11%-16% for patients who received adequate treatment for candidemia, vs. an attributable mortality rate of 31%-41% among patients who did not receive adequate treatment. This would imply a ~60%-65% reduction of the attributable mortality rate due to treatment.
• Cost of treatment (see calculations here):
  • We searched GAFFI’s database for prices of caspofungin, micafungin, and anidulafungin. Unfortunately, prices are available for very few countries, none of which is in Africa. Thus, we collected price data for five LMICs we could find, two of which (Pakistan and Indonesia) are among our top 10 number of IC/C cases (see here). These prices (per daily dose) typically range somewhere in the $100s. We assume (though we haven’t been able to find out) that the prices refer to the medication price only, and do not include other relevant healthcare costs, e.g., for hospitalization.
  • Taking into account hospitalization costs would substantially increase the overall cost estimate, perhaps by ~3-4 times. For example, Neoh et al. (2011) found that, in Australia, almost three-fourths of the treatment costs are due to hospitalization. Moreover, a review study found that in several HICs “the total healthcare cost per patient with IC/C infection was estimated to range from US $48,487 to $157,574, with an average cost of $10,216 to US $37,715 per hospitalization” (Gebretekle et al., 2022; Wan et al., 2020). According to Gebretekle et al. (2022), “patients’ hospital stay is 22–34 days longer compared to those with non-invasive candidiasis.”
  • Treatment cost per case:
    • Our estimated treatment cost per case ranges from $286 to ~$3.6k in the most optimistic scenario (14 days of treatment; lowest treatment price per country).
  • Treatment cost per life saved:
    • In the most optimistic scenario (14 days of treatment; lowest treatment price per country), our estimated treatment cost per life saved ranges from ~$2.6k to ~$33k (depending on the country), but this neither takes into account the cost of necessary hospitalization, nor the cost of diagnosis.
• Barriers to treatment:
  • An echinocandin treatment needs to go through the intravenous route, and individuals need to remain hospitalized (~22-34 days). Our impression is that the treatment is otherwise relatively uncomplicated.
  • Side effects are generally mild (Denning, 2003).
  • “Gold standard” diagnostic tests have low sensitivity and can take too long:
    • Blood culture tests are generally considered the “gold standard” for the diagnosis of invasive candidiasis (Barantsevich & Barantsevich, 2022). Moreover, “the easiest test to diagnose invasive candidiasis is the blood culture test, though the efficiency of the procedure is low: Candida spp. are isolated from blood in only 21–71% of patients with autopsy-proven invasive candidiasis. [...] The main drawback of culture methods is that they are long, with a 72–96 h turnaround time leading to delays in proper treatment that results in increased mortality”. The long time to diagnosis can be problematic, as a “delay in the introduction of antifungal treatment for each 12–24 h may result in increases in mortality rate of up to 100%.”

Appendix H: Estimated R&D spending on fungal diseases by biotech companies

Table I1: Estimated R&D spending on fungal diseases by selected biotech companies

Category	Company	Funding
Antifungal treatments	Pfizer (who acquired Amplyx)	~USD 13M/year90
	F2G [The Rare Fungal Disease Company]	~USD 15M/year
	Scynexis	~USD 11M/year
	Cidara	~USD 8M/year91
Fungal diagnostics	Era Biology	Low confidence guess is around USD 20M/year92
	Richardson Bio-Tech	Not found
Fungal vaccines	Creative Biolabs	Not found
	Biothera	≥USD 150k
	LA BioMed	Not found
	NovaDigm Therapeutics	Not found

Notes

---

1. According to WHO estimates, there are ~620k malaria deaths and ~650k HIV/AIDS deaths worldwide each year. ↩

2. All four sources are directly linked or closely associated with the Manchester Fungal Infection Group. ↩

3. As David Denning (professor of infectious diseases in global health, University of Manchester; formerly a senior advisor at GAFFI) explained, the underestimation is partly due to too few people being tested and partly due to the poor sensitivity of many diagnostic tests, leading to missed cases. ↩

4. The scarcity of DALY estimates has been confirmed by several experts. The only other study estimating fungal disease DALYs we found focuses on invasive aspergillosis in Iran. They estimate 164.13 DALYs per 100k population (Tavakoli et al., 2019), which is roughly aligned with our DALY estimates. ↩

5. See, e.g., Denning et al. (2024, p. 4, Table 2), which shows the presumed “ratio of treated to untreated cases” for individual fungal diseases. ↩

6. The burden of some fungal diseases might be more underestimated than for others. For example, according to Denning, incident cases for histoplasmosis are up to 10x higher globally than reported in the literature. He referred to a study in Nigeria (Oladele et al., 2022) that found that, in many areas, rates of histoplasmosis in AIDS matched or exceeded those of tuberculosis in AIDS. Another example is candidiasis, for which Denning estimates that incident cases are up to 3x higher than normally diagnosed in clinical practice. Invasive candidiasis is usually diagnosed using blood culture (Kullberg & Arendrup, 2015, pp. 1447-1448), but culture diagnostics miss ~50% of invasive candidiasis cases (Clancy & Nguyen, 2013). ↩

7. For example, the WHO (2022) Fungal Priority Pathogens List prioritizes pathogens, not diseases. ↩

8. Pneumocystis pneumonia, where the available estimates suggest the majority of the burden is in Africa, is a notable exception. ↩

9. According to Bongomin et al. (2017), no country estimates of fungal diseases existed for the majority of African countries in 2017 (see Figure 1 in article). Moreover, “comparisons of the incidence across countries is difficult as recent local epidemiological data was only available for 20 of the countries.” ↩

10. This excludes skin, hair, and nail fungal infections, which aren’t typically fatal. ↩

11. According to The Lancet Respiratory Medicine (2016), CPA “almost always presents in people with underlying pulmonary diseases.” ↩

12. Note that this is likely somewhat skewed by different diagnostic efforts for these populations. ↩

13. While no expert volunteered their own estimate, we prompted one interviewee with the idea that the burden could grow by a factor of three to four times. They told us this seemed "“significantly exaggerated", which prompted us to revise our initial assumptions about growth rates slightly downwards. ↩

14. According to Arturo Casadevall (a professor of molecular microbiology and immunology at Johns Hopkins University), the emergence of new fungal pathogens is unpredictable. However, given that in recent decades several new fungal pathogens emerged and there were major disease outbreaks (e.g., Candida auris emerged in 2009; there was an outbreak of black fungus disease in India in 2021 following COVID-19), we expect that this could happen again in the next two decades. ↩

15. >95% of funding is from the public sector (mainly US NIH), and the remainder is from the philanthropic sector (mainly Wellcome Trust). The majority of funding is for basic & early stage research. ↩

16. This guess is based on roughly tripling the G-FINDER project’s estimate of R&D funding for cryptococcal meningitis, mycetoma, and histoplasmosis. ↩

17. Scientific publications in 2017: ~8,800 for tuberculosis, ~5,700 for malaria, 213 for cryptococcosis, 80 for paracoccidioidomycosis, and even less for other fungal diseases (Rodrigues & Albuquerque, 2018, p. 3). ↩

18. While they developed some antifungals in the past (e.g., Novartis developed Lamisil), it is unclear to us whether they are currently developing new antifungals.  ↩

19. According to Rauseo et al. (2020), “[g]reat strides were made in the 1990s … but drug development has largely stalled since then. Currently, 3 main classes of antifungals are approved for the treatment of IFIs: polyenes, azoles, and echinocandins.… The newest class of antifungal drugs, the echinocandins, was discovered in the 1970s and took almost 30 years to be approved.” According to Denning and Bromley (2015, p. 1416), “only four compounds are in active clinical development for the treatment of systemic disease, with a further two agents expected to enter clinical development in 2015.” See Gintjee et al. (2020), Rauseo et al. (2020), and Perfect (2017) for an overview on current antifungal pipeline developments and novel treatments in development. ↩

20. We found out about this at a late stage of this report. Thus, we did not have sufficient time to investigate any concrete actions BARDA is planning. ↩

21. “This is because fungi cells are more closely related to human cells than other microbes such as bacteria. Meaning that compounds toxic to fungi will likely also be toxic to humans” (Albert, 2021). ↩

22. These figures are taken from antibacterial examples, but “antifungals are just as slow,” according to Rex (2022, p. 9). ↩

23. The FKS Valley Fever vaccine was successful in animal trials in the 1970s but found to be not significantly protective in humans during clinical trials in the 1980s (see Amaro & Wood, 2012; Cox & Magee, 2004; Pappagianis, 1993). It was developed by the Valley Fever Vaccine Study Group in California. Subsequent efforts have not reached human trials. ↩

24. The PEV7 (Sap2) vulvovaginal candidiasis vaccine has completed Phase I clinical trials (Sahu et al., 2022). It was developed by now-defunct Pevion Biotech in Switzerland and the Istituto Superiore di Sanità in Italy. The technology was acquired by NovaDigm Therapeutics (De Bernardis, 2018). The NDV-3 (Als3) recurrent vulvovaginal candidiasis vaccine has completed Phase I & II clinical trials (Sahu et al., 2022). It is being developed by NovaDigm Therapeutics in the US. ↩

25. Oliveira et al. (2021) describe such vaccines — which grant broad protection against many fungal species — as the field’s “holy grail.” ↩

26. Oliveira et al. (2021) discuss four candidates tested in mice. A group at the University of Georgia recently demonstrated that the NXT-2 peptide is significantly protective in murine and non-human primate models of invasive aspergillosis, invasive candidiasis, and Pneumocystis pneumonia (Rayens et al., 2022). A university press release said “it could be the first clinically approved immunization to protect against invasive fungal infections” (Beeson, 2023). The US NIAID announced a call for research proposals of vaccines in March 2022, including fungal vaccines, indicating a potential increase in government and industry involvement in the near future (NIAID, 2022a). Among other types of vaccine, the call solicited “[v]accines focused on fungal pathogens, such as but not limited to, Candida auris, other AMR Candida species, Aspergillus fumigatus, Mucorales and Coccidioides” (NIAID, 2022b, p. 7). ↩

27. Several smaller advocacy initiatives exist, e.g., Drugs for Neglected Diseases Initiative (DNDi), International Pediatric Fungal Network (IPFN), International Society for Human and Animal Mycology (ISHAM). We have not reviewed those in detail. ↩

28. See GAFFI’s 2021 Financial Statement (p. 20). ↩

29. A cursory search suggests the London course has likely since shut down, but the University of Exeter has recently opened one. ↩

30. GLASS aims to foster surveillance of antimicrobial resistance and antimicrobial consumption use. GLASS-FUNGI is a global collaboration to collect data on antifungal-resistant infections (WHO, n.d.). ↩

31. Some fungal diseases that are endemic to specific countries/regions have a significant national burden (e.g., Paracoccidioidomycosis is endemic in Brazil and has an incidence of 40/100k inhabitants in some areas [Mendes Peçanha et al., 2022]). ↩

32. A sudden growth in the at-risk population could be triggered by hard-to-predict events that weaken the immune systems of large numbers of people, past examples of which include the HIV/AIDS epidemic and the COVID-19 pandemic. ↩

33. “Clinical suspicion” refers to the awareness and consideration by doctors and health professionals that a fungal infection could be the underlying cause of a patient's symptoms. ↩

34. GAFFI collects and presents information about prices and availability of antifungal drugs across countries in their antifungal drug maps. The underlying study of those drug maps is Kneale et al. (2016).  ↩

35. For example, a 2016 demonstration by GAFFI and the Asociación de Salud Integral in Guatemala showed that HIV patients’ mortality rates from fungal diseases fell by 7 percentage points after major diagnostic improvements (GAFFI, n.d.). Early diagnosis is critical to reducing case fatality rates among immunocompromised patients (Richardson & Warnock, 2022, p. 9). ↩

36. Additional logistical and scientific complications include the high cost of conducting clinical trials with fragile patients (Perfect, 2017) and the diversity of fungal infection sites in human hosts (Oliveira et al., 2021). ↩

37. Our opinion could change depending on the contents of GAFFI’s facilitation model. ↩

38. The model was previously demonstrated in Guatemala and was being undertaken in Argentina when we interviewed GAFFI representatives in early 2023. ↩

39. We don’t trust the finding of a 7 percentage point reduction in mortality in Medina et al. (2021). First of all, our understanding is that this figure comes from a simple before-after comparison, not a randomized controlled trial or other rigorous impact evaluation method. Second, we have some concerns about the statistics reported in the study, as regards the use of “percent” versus “percentage points”. Third, the 7 percentage point difference in mortality between two years is statistically insignificant (p=0.187). We suspect that the study was underpowered to detect such mortality differences. ↩

40. Denning indicated that he unsuccessfully approached Chris Murray, founding director of IHME, around 10 years ago to advocate for the inclusion of fungal diseases in GBD. He suspects his proposal got rejected due to difficulties in disentangling underlying conditions and fungal infections. ↩

41. This is not necessarily a sign of poor quality of sources, but can indicate that the underlying data is so scarce and heterogeneous that extrapolation requires many assumptions which lead to vastly different results. For example, invasive candidiasis has an incidence rate between 2 to 24 per 100k, depending on the country. Extrapolating this to the global level could lead to an estimated number of cases between 400k-4.8M, depending on the assumptions used. ↩

42. We use a schema, slightly simplified from LIFE’s, with the following categories: superficial, chronic lung or deep tissue, allergic, and invasive. ↩

43. An alternative schema, used by the nonprofit group Leading International Fungal Education (LIFE), is broadly similar to that of Richardson and Warnock (2012) but uses the categories “invasive,” which is most similar to “systemic,” and “chronic deep-tissue and lung,” which is somewhat similar to “subcutaneous” (LIFE, n.d.). The Fungal Infection Trust (FIT) uses the term “life-threatening” instead of “invasive” but is otherwise largely aligned with LIFE (FIT, 2020). Note that FIT was the original developer of LIFE and remains closely associated with it (FIT, n.d.). ↩

44. For example, Aspergillus is ubiquitous both indoors and outdoors, yet generally does not cause serious infection. ↩

45. Fungal diseases caused by traumatic implantation are especially common among those who walk barefoot in rural, tropical areas (see Harter, 2022). They are generally regarded as NTDs and include mycetoma, sporotrichosis, and chromoblastomycosis.  ↩

46. For example, Candida lives on most people’s skin and mucous surfaces without causing infection. See Table 1 in Limon et al. (2017) for a list of fungal pathogens by how they interact with mammalian hosts, including whether they are environmentally acquired or commensal. ↩

47. Endemic fungal diseases are also generally distinguished, morphologically, by their ability to switch from yeast to filamental forms when infecting humans; they are “dimorphic,” better able to cause infection — albeit generally not debilitating disease — in immunocompetent individuals, and are thus considered “true pathogens” (Richardson & Warnock, 2012, pp. 6-7). By contrast, other fungal diseases are “opportunistic infections” and generally do not affect the immunocompetent. ↩

48. Misdiagnosis is common in non-specifically presenting systemic fungal diseases. For example, chronic pulmonary aspergillosis is often misdiagnosed as tuberculosis in HIV/AIDS patients (Denning, 2022). ↩

49. Although culture tests are considered the “gold standard” for diagnosis because they help identify the specific pathogen and can be used to test antifungal susceptibility, they are often insensitive (Lass-Flörl, 2017, p. 8). For example, blood cultures only have ~50% sensitivity for diagnosing candidemia and invasive candidiasis (Clancy & Nguyen, 2013), and respiratory tract cultures only have ~50% sensitivity for diagnosing mucormycosis (Lackner et al., 2014). ↩

50. While these are often the most rapid ways to diagnose fungal diseases, serological results “are seldom more than suggestive or supportive of a fungal diagnosis” (Richardson & Warnock, 2012, p. 26). Serum antibodies are ineffective because immunocompromised people do not show robust immune responses to fungal infection. See also Table 3 in Lass-Flörl (2017, pp. 9-10) for a summary of the advantages and disadvantages of such techniques. ↩

51. Molecular tests for fungal diseases are the most recent addition and have contributed to significantly improving the diagnosis of aspergillosis (Denning, 2022). These include PNA FISH and PCR tests. See also Table 4 in Lass-Flörl (2017, p. 13). ↩

52. The four main classes discussed are for systemic treatment. The literature shows variance as to the exact number and composition of the main classes of antifungal drug depending on context. Richardson and Warnock (2012) also include allylamines (used largely as a topical treatment) while lumping flucytosine into a miscellaneous group. In addition, Tom Chiller alluded to the existence of three main classes (likely referring to azoles, echinocandins, and polyenes, which are consistently listed). ↩

53. GAFFI’s list consists of amphotericin B (polyene), flucytosine (pyrimidine analogue), itraconazole (azole), voriconazole (azole), and natamycin (polyene). WHO’s list consists of all of the foregoing except natamycin, plus clotrimazole (azole), fluconazole (azole), griseofulvin (other), and nystatin (polyene). ↩

54. Concerningly, they found that intravenous deoxycholate amphotericin B and flucytosine were unavailable in 27% and 76% of countries, respectively, and their prices varied from under USD 1 to USD 171 and from USD 4.60 to USD 1,409, respectively. ↩

55. In technical terms, antifungal resistance is defined as “the ability to grow at antifungal drug concentrations above a defined antifungal susceptibility break point … expressed as a minimum inhibitory concentration (MIC)” (Fisher et al., 2022, p. 557). Relatedly, drug tolerance occurs when susceptibility is diminished such that cells grow more slowly at or above MICs, and is most relevant for fungistatic — i.e., growth-inhibiting — drugs. ↩

56. In vivo resistance is a particular issue for azoles and echinocandin, while environmental resistance largely affects azoles. ↩

57. This refers to the number of new cases of a disease each year. ↩

58. Rodríguez Tudela pointed to a study which showed that histoplasmosis mortality has decreased from 33% to 21% from 2017 to 2019, which he said “was attributed to the implementation of advanced diagnostic methods that enabled the swift identification of fungal opportunistic infections in individuals living with HIV” (Medina et al., 2021). He further mentioned that the study was conducted in Guatemala but that parallel initiatives to increase access have also been undertaken in Buenos Aires and Santa Fe, Argentina. We don’t know whether the mortality reduction can be extrapolated to other diseases, but we assume so to remain conservative. It also seems plausible that mortality rates could be upwardly biased by a selection bias of underdiagnosis of cases and sicker people more likely to be diagnosed. ↩

59. We multiplied our estimated annual number of deaths by the counterfactual life expectancies of those who have died. For simplicity reasons, we assumed the same counterfactual life expectancy across diseases (7-25 years [80% CI]) for almost all fungal diseases (except the top two in terms of mortality, i.e., chronic pulmonary aspergillosis and candidemia/invasive candidiasis). Our assumed counterfactual life expectancy (for all but the top two diseases) was a rough guess, informed by several factors: 1) Most fungal diseases affect all age groups (i.e., not specifically very old or very young people). Thus, our starting point was Open Philanthropy’s default assumption of 32 YLLs for individuals aged >5 years. 2) Many severe fungal disease cases occur in people with comorbidities (e.g., HIV/AIDS, tuberculosis), which means that they have a shorter life expectancy than the average population. 3) Our lower bound of seven years seemed conservative to us, as many people with underlying conditions have a close-to-normal life expectancy if they are diagnosed in time and are treated.

    We then spent ~30 mins thinking about counterfactual life expectancies for the top two diseases in terms of mortality, as these have by far the largest contribution to total deaths (and therefore DALYs). Chronic pulmonary aspergillosis: ~90% of cases are in patients with respiratory diseases (e.g., tuberculosis and COPD). Given the typical life expectancies and mortality rates of those patients, our best guess is that the counterfactual life expectancy is roughly in line with our assumed average life expectancy for the other fungal diseases (i.e., seven to 25 years [80% CI]). Invasive candidiasis and candidemia: We expect counterfactual life expectancy to be a bit lower relative to the average fungal disease, as 60% of cases are in ICU patients, which typically have very high mortality rates (even higher than other immunocompromised/ill patients). Based on this observation, our rough guess is that the counterfactual life expectancy is ~20% lower than for the average fungal disease, i.e., ~5.6-20 years (80% CI). ↩

60. According to Denning et al. (2022), “the mortality rate could be inflated by selective reporting of hospitalized cases or those at tertiary care centres, which would tend to see the worst cases, and by patients older than the usual age of tuberculosis onset in many parts of the world, including India. On the other hand, the lack of a CPA diagnosis, and either no treatment or mistreatment, is likely to enhance mortality.” ↩

61. To our understanding, recent prevalence studies (e.g., Denning et al. (2022) for India) do not clearly distinguish between hospitalized and nonhospitalized patients, and we expect that they are a combination of both. ↩

62. Lowes et al. (2017) reviewed 10 studies and Denning et al. (2022) reviewed 26 studies. ↩

63. According to Rodríguez Tudela and Denning, there are very few cohort mortality data published from resource-poor countries. ↩

64. Rodríguez Tudela said that, in high-income countries, clinicians correctly identified invasive aspergillosis as the cause of death in only 27% of fatal invasive aspergillosis cases identified at autopsy. ↩

65. They raised the following example: “[L]et us take invasive candidiasis and Candida bloodstream infection. The population incidence rate per year of bloodstream Candida infection is between 2 and 24/100,000. Lower in countries with excellent antibiotic control and few critically ill patients and highest in middle income countries with poor infection control and unbridled antibiotic use. So, if a global population of 8 billion, there will be between 160,000—1,920,000 cases. BUT blood culture is only 40% sensitive, and the figures are taken from positive blood cultures. PCR and glucan are better, so then if we multiply by 2.5 to reach 100%, the figures would be = 400,000 — 4,800,000 invasive candidiasis cases. Huge range as you can see. Without good surveillance nothing good can be expected.” ↩

66. See cell “V26” in this sheet. We calculated this as a rough replication and extension with newer data of Bongomin et al. (2017), as we explained here, but many countries are omitted from this analysis. ↩

67. Numerical estimates in Bongomin et al. (2020) and Denning and Chakrabarti (2021) about frequency of underlying conditions derive from one HIC (UK) patient sample (see Smith & Denning, 2011). Newer TB co-infection data appear to exist for LMICs — Indonesia (Setianingrum et al., 2022) and Nigeria (Oladele et al., 2017) — but TB patients are diagnosed with CPA, not the other way around. ↩

68. Information about frequency of underlying conditions could depend heavily on three studies conducted on HIC populations: the US (Atlanta and Baltimore metro areas; Cleveland et al., 2015), France (Paris metro area; Lortholary et al., 2014), and Denmark (Arendrup et al., 2011); see citations for Table 1 in Kullberg and Arendrup (2015, p. 1447). ↩

69. “Patients are usually not immunocompromised by HIV-infection, cancer chemotherapy or immunosuppressive therapy” (Denning et al., 2016, p. 51). ↩

70. While mucocutaneous forms of candidiasis are common in HIV-infected patients, HIV infection has no apparent association with invasive candidiasis and candidemia (Ampel, 1996, p. 110; Limper et al., 2017, p. e334). ↩

71. It is additionally estimated that ~5% of CPA patients have “lung irradiation,” which may include present or prior radiation therapy for lung cancer (Denning & Chakrabarti, 2017, p. e359, Table 1). Previous lung cancer is widely thought to be a risk factor (see, e.g., Denning et al., 2016, p. 55; Denning & Chakrabarti, 2017, p. e358, Figure 1; Bongomin et al., 2020, p. 4), but numerical estimates of risk factor frequency are not available. ↩

72. See Table 1 in Kullberg and Arendrup (2015, p. 1447). ↩

73. See Table 1 in Denning and Chakrabarti (2017, p. e359). “Impairment of lung function with … aspergillomas and chronic pulmonary aspergillosis are known complications and are more frequent in patients with drug-resistant TB than in patients with drug-sensitive TB” (Pai et al., 2016, p. 17). ↩

74. See Chen et al. (2015) and Fontalvo et al. (2016). ↩

75. “The vast majority of patients with CPA have underlying structural lung diseases” (Bongomin et al., 2020, p. 4). In particular, the most common underlying lung disease appears to be emphysema (especially bullous emphysema) or COPD, estimated to be present in ~30%-50% of CPA patients (Denning & Chakrabarti, 2017, p. e359, Table 1). Note also that the Fungal Infection Trust (2020) seems to imply that effectively 100% of the global burden of CPA falls under the “Respiratory” category of underlying disease group (p. 8). ↩

76. The Fungal Infection Trust (2020) estimates that ~600k out of ~950k (63%) annual cases are in patients with “Immune deficit” and ~300k cases (32%) are in patients in “Critical care,” implying the remaining ~50k do not have a risk factor that falls under the available categories (p. 8). ↩

77. See Fungal Infection Trust (2020, p. 8). ↩

78. Note that we select these weights somewhat arbitrarily, using Table D1 as a starting point. See cell notes in column C here for our reasoning. ↩

79. “70% of individuals with sputum smear-positive pulmonary TB died within 10 years of being diagnosed” (World Health Organization, 2022, p. 37). See also Tiemersma et al. (2011). ↩

80. See Hoger et al. (2016) and Moosadazeh et al. (2014). ↩

81. Note: “N/A” indicates that we have not been able to find any information in a quick literature search. We expect that 10 more hours of research could help fill significant gaps in the table, though we also expect that on some aspects no or very limited data is available. ↩

82. For example, in Figures 4-5, country-level case rates are averaged for each continent but the averages are not weighted by country population. In Figures 2-3, raw case numbers are averaged for the countries in each continent; the average does not seem to be a meaningful statistic. ↩

83. We took estimates from a number of individual studies published since 2017 that GAFFI has collated, prioritizing the countries with the largest populations as we were short on time. Data for Japan, Australia, and New Zealand were separately sourced from research posters and conference abstracts from the Aspergillus & Aspergillosis website. ↩

84. The country-level estimates are for candidemia, but we assume that candidemia incidence rates generally track those of invasive candidiasis more broadly. ↩

85. This is the first international research center focused on fungal infections. ↩

86. A common measure of treatment effectiveness for CPA is a quality of life measure specific to respiratory diseases, the St. George’s Respiratory Questionnaire (SGRQ). We decided to omit those studies as it’s not straightforward to interpret and to translate into DALYs or mortality. ↩

87. There seems to be no uniform definition of what “response” means in the context of CPA. Our impression is that it usually refers to some kind of measure of whether symptoms have improved. This can be both subjective, but also measured by physiological markers. It does not provide information on whether a person was completely cured. ↩

88. In a later follow-up in 2024, Denning added that he is “in the midst of a new assessment of mortality.” ↩

89. This is unlikely to be accurate for various reasons, for example: (1) The mortality rate of 15% mentioned in the literature is based mostly on treated individuals (see Denning et al., 2011). We have not found any mortality figures for untreated CPA patients; (2) Given that CPA can rarely be fully cured, it may be that deaths are not really averted, but just delayed by some months or years; (3) Given high side effects, we expect that a substantial number of patients may find it difficult to adhere to the treatment, but we don’t know how many. ↩

90. This refers to Amplyx’s estimated yearly funding before it was acquired by Pfizer. In a quick search, we didn’t find Pfizer’s current R&D spending related to fungal diseases. ↩

91. Assuming half of its funding is for fungal diseases. ↩

92. This guess is not based on any concrete funding figures, but rather on our impression that they are a comparatively large player in fungal disease R&D based on their products and activities. ↩