Article Text

Download PDFPDF
MDM2 gene amplification as selection tool for innovative targeted approaches in PD-L1 positive or negative muscle-invasive urothelial bladder carcinoma
  1. Matteo Brunelli1,2,
  2. Alessandro Tafuri3,
  3. Luca Cima4,
  4. Maria Angela Cerruto3,
  5. Michele Milella5,
  6. Andrea Zivi5,
  7. Sebastiano Buti6,
  8. Melissa Bersanelli6,
  9. Giuseppe Fornarini7,
  10. Valerio Gaetano Vellone7,
  11. Sara Elena Rebuzzi7,
  12. Giuseppe Procopio8,
  13. Elena Verzoni8,
  14. Sergio Bracarda9,
  15. Roberto Sabbatini10,
  16. Cinzia Baldessari10,
  17. Albino Eccher1,
  18. Rodolfo Passalacqua11,
  19. Bruno Perrucci11,
  20. Maria Olga Giganti11,
  21. Maddalena Donini11,
  22. Stefano Panni11,
  23. Marcello Tucci12,
  24. Veronica Prati13,
  25. Cinzia Ortega13,
  26. Anna Caliò1,
  27. Filippo Alongi14,
  28. Enrico Munari15,
  29. Giovanni Pappagallo16,
  30. Roberto Iacovelli17,
  31. Alessandra Mosca18,
  32. Camillo Porta19,20,
  33. Guido Martignoni1,21,
  34. Alessandro Antonelli3
  1. 1 Department of Diagnostics and Public Health, Pathology Unit, University and Hospital Trust of Verona, Verona, Italy
  2. 2 Department of Diagnostics and Public Health, FISH Lab, University and Hospital Trust of Verona, Verona, Italy
  3. 3 Division of Urology, University and Hospital Trust of Verona, Verona, Italy
  4. 4 Department of Clinical Services, Pathology Unit, Santa Chiara Hospital, Trento, Italy
  5. 5 Division of Oncology, University and Hospital Trust of Verona, Verona, Italy
  6. 6 Division of Oncology, University and Hospital Trust of Parma, Parma, Italy
  7. 7 Division of Oncology, San Martino Hospital, Genova, Italy
  8. 8 Division of Oncology, IRCCS Foundation, Istituto Nazionale dei Tumori di Milano, Milano, Italy
  9. 9 Division of Oncology, Santa Maria Hospital, Terni, Italy
  10. 10 Department of Oncology, Hematology & Respiratory Diseases, Division of Oncology, University of Modena & Reggio Emilia, Modena, Modena & Reggio Emilia, Italy
  11. 11 Division of Oncology, Hospital Trust of Cremona, Cremona, Italy
  12. 12 Division of Oncology, Cardinal Massaia Hospital, Asti, Italy
  13. 13 Division of Oncology, Institute for Cancer Research and Treatment, Asl Cn2 Alba-Brà, Alba-Brà, Italy
  14. 14 Division of Radiotherapy, Ospedale SacroCuore di Negrar di Valpolicella, Negrar, Italy
  15. 15 Division of Pathology, Ospedale SacroCuore di Negrar di Valpolicella, Negrar, Italy
  16. 16 Clinical Epidemiologist, Silea, Italy
  17. 17 Policlinico Universitario Fondazione Andrea Gemelli, Rome, Italy
  18. 18 Oncology, Candiolo Cancer Institute, IRCCS-FPO, Torino, Italy
  19. 19 Department of Biomedical Sciences and Human Oncology, University of Bari ‘A.Moro’, Bari, Italy
  20. 20 Division of Medical Oncology, Policlinico Consorziale di Bari, Bari, Italy
  21. 21 Pathology Unit, Pederzoli Hospital, Italy
  1. Correspondence to Professor Matteo Brunelli, Department of Diagnostics and Public Health, Pathology Unit, University and Hospital Trust of Verona, 37134 Verona, Italy; matteo.brunelli{at}


Aims According to The Cancer Genome Atlas (TCGA), around 9% of bladder carcinomas usually show abnormalities of the murine double minute 2 (MDM2) gene, but a few studies have been investigated them. We profiled MDM2 gene amplification in a series of urothelial carcinomas (UC) considering the molecular subtypes and expression of programmed death ligand 1 (PD-L1).

Methods 117 patients with muscle-invasive UC (pT2-3) without (N0) or with (N+) lymph-node metastases were revised. Only cases with availability of in toto specimens and follow-up were studied. Tissue microarray was built. p53, ER, RB1, GATA-3, CK20, CK5/6, CD44 and PD-L1 (clone sp263) immunoexpression was evaluated. Fluorescent in situ hybridisation was assessed by using the HER-2/neu, FGFR-3, CDKN2A and MDM2 probes. True (ratio 12q/CEP12 >2) MDM2 gene amplification was distinguished from polyploidy/gains (ratio <2, absolute copy number of MDM-2 >2). MDM2 and PD-L1 values were correlated to the TCGA molecular phenotypes. Statistical analysis was performed.

Results 6/50 (12%) cases (5 N0 and 1 N+) were amplified for MDM2 without matching to molecular phenotypes. Of 50, 14 (37%) cases expressed PD-L1 at 1% cut-off; 3/50 (9%) at >50% cut-off; of these, 2 cases on side of neoplasia among inflammatory cells. Only one out of six (17%) cases amplified for MDM2 showed expression (>50% cut-off) of PD-L1. MDM2 amplification was independent to all documented profiles (k test=0.3) and was prevalent in recurrent UC.

Conclusion MDM2 amplification has been seen in both PD-L1 positive and negative muscle-invasive bladder UC independently from the TCGA molecular phenotypes. MDM2 and PD-L1 might be assessed in order to predict a better response to combo/single targeted therapies.

  • Urinary Bladder
  • Pathology
  • Molecular

Data availability statement

No data are available.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Several molecular classifications have been proposed to stratify bladder carcinoma,1 2 however, their heterogeneity precludes their widespread clinical application in everyday clinical practice, usually due to absence of simultaneous morphological and molecular analysis. Murine double minute 2 (MDM2) gene amplification is reported in tumours such as glioblastoma3 and retroperitoneal sarcomas,4 characterised by aggressiveness and chemoresistance. Moreover, the pathway involving programmed death protein 1 (PD-1) and its ligand (PD-L1) is a well-characterised immune checkpoint and has been applied in the clinical treatment of various cancer,5 being also studied in precancerous lesions.6

In the setting of bladder carcinoma, little is known about MDM2 genomic abnormalities that cover around 9% of bladder carcinoma according to The Cancer Genome Atlas (TCGA) reports.1 Moreover, this molecular phenotype has not been correlated yet with expression of PD-L1 in urothelial carcinoma (UC) and/or with luminal/basal/other molecular phenotypes.7 8 Notably, basket trials with Mdm2 inhibitors have been already designed, thus paving the way to the use of Mdm2 targeting therapies in tumours characterised by MDM2 amplification in order to improve outcomes. Systemic platinum-based combinations are still the cornerstone of UC treatment, with overall response rates close to 50%. However, these responses are short-lived, with a median progression free survival of approximately 8 months.9 Furthermore, platinum-based chemotherapies are related to a high incidence of adverse events and many patients are not-eligible for cisplatin (the most active platinum compound in this setting), because of a number of hindering factors.10 Recent discoveries about the tumour microenvironment lead to the development of immune check point inhibitors (particularly for PD-L1 expressing tumours), which proved to be a valid alternative for metastatic bladder cancer patients who are not platinum-eligible, as well as after failure of previous platinum-based therapy.11 12 Additionally, in the last few years, many genetic characteristics have been studied as possible therapeutic targets; among these, MDM2 in bladder cancer has recently garnered great interest.13 14 The advent of molecular-based therapies has revolutionised the therapeutic approach to different cancer by focusing the attention on various targets. The discovery of the negative feedback regulation mechanism between Mdm2 and p53 proteins could be considered a promising way to restore the normal oncosuppressive function of p53.15 In this context, MDM2 gene expression needs to be additionally assessed.

The aim of this project is to evaluate the MDM2 gene expression and its correlation with both PD-L1 expression, as well as molecular (luminal/basal/others) phenotypes in a series of pT2-3 bladder UCs without (N0) and with (N+) lymph node metastases and to determine its possible application as a biological marker.

Materials and methods

Informed consent was obtained in writing from living patients or relatives for all tissues used in this study from the Urology Clinic.

Case selection and tissue sampling

Tissue obtained from patients with pT2-3 bladder UCs who underwent radical cystectomy (figure 1) and regional lymphadenectomy, was processed. Follow-up was recorded. The cases were registered at the Pathology Unit of the AOUI Hospital Trust of Verona.

Figure 1

Example of invasive bladder urothelial carcinoma; invasion of the suburothelial lamina propria, T1 (A); invasion of the superficial muscle layer, T2a (B); H&E, ×20.

Pathological tissue was fixed in buffered formalin and paraffin embedded. Then, tissue samples were sectioned into slices with a thickness of 3–5 µm using a microtome and were assembled on a polarised slide.

Tissue microarrays

A tissue microarray (TMA) was built after selecting representative areas of each tumour specimen, with array cores measuring 2 mm maximum in diameter (figure 2A). Three tumour cores were obtained for each case.

Figure 2

Tissue microarray built per bladder carcinoma: examplification of the punched tissue cores (A); sections of formalin-fixed and paraffin-embedded tissue from which tissue microarray was obtained (B).

Immunohistochemical analysis

Immunohistochemistry was carried out using an automated Ventana Systems or Leica Bond for markers associated with luminal (GATA3, CK20 and ER) and basal phenotype (CK5/6 and CD44) and p53/RB1. Immunostaining was performed on formalin-fixed paraffin-embedded sections using complete automated IHC platform (Benchmark). Tissue sections were deparaffinised and hydrated, and heat-induced epitope retrieval was performed for each stain. After cooling and rinsing sections with reaction buffer, primary antibody was applied. Signals were detected with HRP labelled secondary antibody, followed by DAB chromogen as per manufacturer’s instructions. Percentages of neoplastic cells were scored and positive tumours were assessed when at least 10% of cells displayed positive expression.

Immunohistochemical analysis for PD-L1 expression

Stained TMA sections were scanned using Olympus BX and scored on the basis of the percentage of tumour cells showing membranous positivity, irrespective of staining intensities; a three-tiered system was then applied using the following thresholds: <1%, ≥1% to <50% and ≥50%. For the discordant cases, we evaluated cases simultaneously (MB and GM) and interpreted after sum up of the overall three cores.

PD-L1 evaluation has been performed blindly by two experienced uropathologists trained for PD-L1 evaluation, designated pathologist A (MB) and pathologist B (GM), respectively, who routinely use the SP263 clone.

Placenta was used as external control, whereas macrophages were used as internal control, in order to validate the adequacy of PD-L1 staining reaction.

Fluorescent in situ hybridisation

The TMA technique in conjunction with fluorescent in situ hybridisation (FISH) was used to detect the amplification of the MDM2 gene at the chromosomal level on the white TMA slices of selected cases. This analysis was carried out on interphase nuclei of sections of formalin-fixed and paraffin-embedded tissue from which TMA was obtained, with a thickness of 3 µm that were mounted on positively charged slides (Dako, Glostrup, Denmark) (figure 1B). The Kreatech commercial FISH probe MDM2 (12q15)/SE12 (Leica Biosystem, Amsterdam, Netherland) was used for the diagnostic use and detection of the chromosomal region of the human MDM2 gene and of the alpha-satellites of chromosome 12 by FISH.

EGFR (Abbott spa, Milan), CDKN2A (locus chromosome 9 p) (Abbott Spa, Milan), Her-2/neu (Abbott Spa, Milan) and FGFR2-3 (ZytoLight, ZytoVision, Bio-Optica) were also tested.

Molecular interpretation

The combined probe allowed the detection of both the increase in the number of copies (copy number gain) of the chromosome, as well as the amplification of the MDM2 gene (LSI/CEP >2 count ratio). The results were evaluated with an Olympus BX61 fluorescence microscope (Olympus, Hamburg, Germany) equipped with filters specific for the fluorochromes used: SpectrumGreen, SpectrumOrange and DAPI. The signals at the region of interest of the MDM2 gene appear red while the signals related to the centromere of the chromosome appear green. After preparing the TMA grid, we assessed the various cores by counting the signals within each core as follows: (1) the core was defined as not amplified when there were two green signals for CEP12 and two red signals for MDM2; (2) the nucleus containing ≥6 red spots was considered amplified for the gene; (3) at least 50 nuclei were present in each core; (4) the nuclei that demonstrate excessive disintegration were excluded; (5) two signals of the same size separated by a distance less than or equal to the diameter of a single signal were considered as a single signal; (6) in the presence of MDM2 gene clusters, the exact number of signals could not be counted and were estimated; (7) if the ratio between the MDM2 and CEP12 signals was 10% of the observed nuclei, the tumour was considered positive for the amplification of the MDM2 gene; (8) in the presence of an elevated number of signals of both MDM2 and CEP12, it was evaluated as copy number gain for both MDM2 and CEP12. For the discordant cases, we considered the most represented MDM2 gene status (amplification, gains/polyploidy, disomy) after sum up of the three cores.

Gene copy number per assessed per probe and corrected per ratio (LSI/CEPs) also per 9p chromosome (CDKN2A). Gene copy number per assessed per FGFR2-3 probes and amplification of HER-2/neu was also assessed by using the AIOM-SIAPEC protocol.

Statistical analysis and clinical data

The kappa-statistic (Cohen’s) measure of interrater agreement (MDM2 gene amplification, PD-L1 expression versus molecular phenotypes (luminal papillary, luminal non-specified, luminal unstable, stroma-rich, basal/squamous and neuroendocrine-like UCs) was calculated by the use of Stata software Version 16 (StataCorp). MDM2 amplification and PD-L1 expression were matched with clinical outcomes.


From 117 pT2-3 patients with bladder UCs, 37 N0 cases were recruited as first cohort. Of these cases, 24 patients were male, 13 were female and the median age was 76.5 (range 56–88) years. Overall, 8 cases were luminal papillary, 3 luminal non-specified, 6 luminal unstable, 6 stroma-rich, 13 basal/squamous and 1 neuroendocrine-like UCs (table 1).

Table 1

Clinical, histopathological and molecular features of case series with MDM2 and PD-L1 status

Additionally, 13 cases with lymph-nodal metastases (N+) were recruited as second cohort. All tissue samples were included in the revision and fixed in buffered formalin and paraffin embedded. Of these cases, seven patients were male, six were female and the median age was 75 (range 69–88). Overall, three cases were luminal papillary, one luminal non-specified, two luminal unstable, two stroma-rich, four basal-squamous and one neuroendocrine-like UCs (table 1).

Immunophenotypical findings

Most cases lacked ER expression (83%) and were GATA3 positive (87%). CK5/6, CD44 and CK20 were positive in 32%, 45% and 37%, respectively. p53 was observed in 6% of cases.

At a≥1% cut-off, the percentage of PD-L1 (sp263) positive cases for both uro-pathologists (MB and GM) were 37% (14 cases), without association with age and gender. In 10/14 cases (71%), the tissue was obtained from primary bladder UCs; in the remaining four cases (29%), from lymph node metastases. At a≥50% cut-off, PD-L1 (sp263) expression was registered in 9% of the cases (three cases), without association with age and gender. Two out of three (67%) were primary bladder UCs and 1 (33%) was a lymph node metastasis. Notably, 2/17 cases (12%), displayed a PD-L1 immunoexpression at ≥50% cut-off on the side of neoplastic nodules among inflammatory cells on the edge of neoplastic foci (figure 3). A skewing predominance of T2 (9, 53%) vs T3 (8, 47%) was found among cases with PD-L1 expression, with 12 N0 and 5 N+ bladder UCs.

Figure 3

Immunoexpression of PD-L1 (sp263) in bladder carcinomas: strong diffuse expression (>50%) of neoplastic cells (immune enrichment) (A); absence of immunoexpression (immune-desertification) (B); heterogeneous positive expression ranging from 5% to <50% of whole neoplastic foci (low-heterogeneous immunoenvironment) (C); absence of expression on neoplastic cells with inflammatory cells at the edge of the neoplasia (immune attack) (D). PD-L1, programmed death ligand 1.

We observed that PD-L1 expression was independent to all documented molecular phenotypes (k test=0.3).

Molecular findings

MDM2 gene amplification was observed in 6/50 cases (12%), without association with age and gender. Interestingly, positive samples with MDM2 gene amplification derived almost only from the primary bladder UC, in fact among the cases with lymph node metastases, MDM2 gene amplification was observed only in 1 out of 13 (8%) cases. A predominance of T2 cases (4/6, 67%) was found among cases with amplification, with 3 N0 and 1 N+ bladder UCs.

Overall, 5 cases out of 50 (10%) were identified as harbouring MDM2 copy number gains/polyploidy, where both gain of signals for MDM2 and for CEP12 internal control were seen. The remaining cases appeared not amplified, showing two signals for MDM2 and two for CEP12 (disomy). The FISH MDM2/FGFR2-3 gene status is showed in figure 4. Her-2/neu gene amplification was positive in 32% of cases.

Figure 4

MDM2 and FGFR-2 gene assessment in bladder carcinoma: two cases with amplification using fluorescent in situ hybridisation technique. The amplified region of the MDM2 gene appears red, while the signals related to the centromere of the chromosome appears green (control). Ratio >2 shows high level of MDM-2 gene amplification (granular pattern) (A); low level gene MDM-2 amplification (B). FGFR-2 gene copy number abnormalities (C) and wild status (two copies of the locus specific gene) (D). MDM-2, murine double minute 2.

FGFR2-3/CDKN2A genomic FISH alterations (mainly gains/amplification) were enriched in 83% of luminal-papillary bladder UCs; p53/HER-2 amplification characterised 83% of luminal unstable bladder UCs; p53 and RB1 characterised the neuroendocrine-like bladder UCs.

Only one case out of six (17%) harbouring MDM-2 gene amplification showed a diffuse expression (at ≥50% cut-off) of PD-L1. Two out of six (33%) expressed PD-L1 ≥1x<50%, the remaining 3/6 (50%) of cases did not express PD-L1. Moreover MDM-2 gene amplification did not match with luminal or basal/other molecular phenotypes.

We observed that MDM2 gene amplification was independent to all documented molecular phenotypes (k test=0.3). Moreover, although it was a relatively uncommon event in this tumour (12% of cases), it occurred after mean intervals of 24 months (range 5–36 months, overall 886 months of follow-up). Notably, 4/6 (66%) of cases with MDM2 gene amplification were recurrent UCs (after an interval of 12, 14, 15 and 19 months, respectively). This finding suggests a possible role of MDM2 gene amplification as independent predictor of tumour relapse.


Genomic taxonomy in muscle invasive bladder cancer was first introduced by TCGA.1 The MDM2 gene anomalies were described to be observed in a small subset of cases (9%), with few analysis on series of bladder carcinoma and little attention onto their potential clinical relevance. In the present study, we showed that among 50 cases of pT2-3 bladder UCs (37 N0 and 13 N+), 12% of pT2-3 N0 and 17% of pT2-3 N+, displayed an MDM2 gene amplification, irrespective to the molecular TCGA classic profiling such as luminal papillary, luminal nonspecified, luminal unstable, stroma-rich, basal/squamous and neuroendocrine-like UCs and, notably, irrespective of the value of PD-L1 (positive/negative immunoexpression).

At the biological level, amplifiers (MDM2 gene amplification) were included according to strictly interphase gene amplification criteria such as using the ratio MDM2 locus specific gene/centromeric control probe CEP12 >2, with a further 10% of cases showing only copy number gains with both gains of MDM-2 and CEP12 with ratio <2 without real amplification, thus showing polyploidy/gains characters. Notably, only one case out of six (17%) harbouring MDM2 gene amplification showed diffuse expression of PD-L1 (sp263), whereas 2/6 (33%) expressed ≥1x<50% and the remaining 50% of cases did not express PD-L1. Our findings promote the evaluation of MDM2 gene amplification in both PD-L1 positive and negative carcinoma in pioneering clinical trials.

Overall, more than 150 000 patients die of bladder carcinoma each year.16 17 At a morphological level, 95% of the cases are UC and the other 5% are squamous cell carcinomas, adenocarcinomas, sarcomas, small cell carcinomas.18 Treatment options depend on the stage of the cancer and may include some combinations of surgery, radiation therapy, chemotherapy or immunotherapy.9 12 19 20 Detailed surgical options may include transurethral resection, partial or complete removal of the bladder or urinary diversion.

MDM2 was initially discovered with its oncogenic potential in the amplified murine cell line 3T3-DM.21 It was postulated that one or more of these genes, present on stable double-minute fragments, conferred a growth advantage to cultured cells and contributed to their tumorigenic potential in vivo, as well as thymus-deprived mice, in which over-expression of the MDM2 gene but not of MDM1 or 3, resulted in tumour growth. Findings from Simon et al 22 reported a percentage of amplified cases of 5.8%, lower than in our analysis. Possible explanations may be due to differences in patient selection or differences in FISH methodology.

According to our findings, MDM2 can be considered a good biological marker having a specific correlation with the local control of the tumour and a precise determination of MDM2 gene amplification can be carried out in a short time by using common FISH analysis on formalin-fixed material. Thus, the presently ongoing development of investigational drugs directed against Mdm2 protein, and in particular its link with p53, renders these compounds extremely intriguing for their therapeutic potential.

Mutational analyses showed that differences between muscle invasive and metastatic bladder UC contribute to the variability in response to chemotherapy and immunotherapy, the heterogeneity does occur at different levels and these alterations may be clinically relevant.7 Profiling morphology-based bladder cancer with molecular analysis will assess identification of the extent of heterogeneity which might change the multimodal treatment of this disease, in order to target different drivers.2

The MDM2 gene, is located in the 12q14.3-12q15 region, and encodes an E3 ubiquitin ligase protein; furthermore, MDM2 activity is related to p53 inhibition, and MDM2 overexpression alters p53 controlled arrest and apoptotic responses to cellular insults, leading to an uncontrolled cell cycle progression and proliferation.23

Within the literature, no studies matched the information resulting from both MDM2 gene assessment and PD-L1 overexpression, or stratified per molecular phenotypes such as luminal papillary, luminal non-specified, luminal unstable, stroma-rich, basal/squamous and neuroendocrine-like UC.

According to our findings, may be of interest to analyse by FISH the amplification status of the MDM2 gene in subtypes of bladder UC in order to eventually treat patients with binding inhibitory molecules especially in cases with localised muscle-invasive tumour who underwent to neoadjuvant chemotherapy or chemoimmunotherapy approaches. Restoring the oncosuppression by the p53 pathway may result in improved therapeutic outcomes. Even more, clinical trials may theoretically consider clustering advanced bladder UC with both on/off MDM2/PD-L1 activation, once again irrespective of luminal or basal molecular/other phenotypes.

Both MDM2 and PD-L1 should be assessed to better characterise luminal papillary, luminal non-specified, luminal unstable, stroma-rich, basal/squamous and neuroendocrine-like UCs, in order to improve clinicopathological correlation and/or predictiveness to targeted therapies such as Mdm2 chemoinhibitors or immunotherapies targeting the PD1/PD-L1 pathway. Combining the profiling of immune-enrichment (characterised by PD-L1+) versus immune-desertification (by absence of PD-L1) versus immune-attack in/to bladder carcinoma (strong PD-L1 expression on the edge of the neoplastic nodules), with the level of MDM2 gene amplification may be useful to select patients at tissue level, to predict response to combo/single targeted therapies. In this context MDM2 gene amplification and gains of MDM2 are different biological patterns.

Pioneering prospective clinical trials with larger patient populations and additional stratification factors are claimed to test innovative combinations vs more consolidated scheme of treatment in muscle-invasive bladder cancer.

Take home messages

  • Advanced bladder urothelial carcinoma may show MDM2 gene amplification.

  • MDM2 gene amplification can be found in both PD-L1 positive and negative bladder urothelial carcinomas.

  • MDM2 gene amplification can be an independent predictor of tumour relapse.

  • Simultaneous MDM2/PD-L1 molecular testing can predict a better response to combo/single targeted therapies.

Data availability statement

No data are available.

Ethics statements

Patient consent for publication

Ethics approval

This study was approved by the institutional review board of the University of Verona in accordance with the Declaration of Helsinki of 1975.



  • MB and AT are joint first authors.

  • GM and AA are joint senior authors.

  • Handling editor Runjan Chetty.

  • Twitter @AlessandraMosc5

  • Contributors All authors contributed equally to all stages of the work.

  • Funding Internal funding from Department of Diagnostics and Public Health (MB, FUR 2018-2019) has been used in part for study-related facilities.

  • Disclaimer This study did not receive any funding from the National Institutes of Health (NIH), the Wellcome Trust and the Howard Hughes Medical Institute (HHNI).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.