Abstract

Background: Previous studies reported the mutational landscape in extramammary Paget’s disease (EMPD); however, the prognostic implications of genetic alterations remain unexplored. While CDKN2A loss is known to be associated with tumor progression or poor prognosis in some types of cancer, its significance in EMPD has not been investigated.

Objectives: To examine the association between common genetic alterations and prognosis in EMPD.

Methods: This is a retrospective cohort study analyzing EMPD cases registered until January 2024 in the Center for Cancer Genomics and Advanced Therapeutics database, which is a nationwide database recording clinical data and comprehensive genomic profiling (CGP) test results in Japan.

Results: A total of 167 cases were recorded in the database, with CDKN2A loss being the most frequent genetic variant. Survival analysis was conducted on 127 cases. Survival from chemotherapy initiation was analyzed with adjusting for length bias inherent in the database using the Kaplan-Meier estimator, an established adjustment method. Cases with BRCA2-mutant tumors (n=18) had a worse prognosis than those with BRCA2-wild-type tumors (n=109; HR=2.97, 95% CI 1.46-6.01, p=0.003). Additionally, CDKN2A-mutant group (n=72) had a significantly worse prognosis than those with CDKN2A-wild-type group (n=55; HR=1.81, 95% CI 1.06-3.07, p=0.029). Most CDKN2A variants were pathogenic, primarily characterized by loss, while most BRCA2 variants were variants of uncertain significance. In the analysis of survival from CGP enrollment based on Eastern Cooperative Oncology Group performance status (ECOG-PS), cases with ECOG-PS 1 at the time of CGP enrollment had significantly poorer prognosis than those with ECOG-PS 0 (p=0.034; median survival time, 531 vs. 259 days).

Conclusions: Somatic CDKN2A variant, mainly exhibiting loss, may be associated with poor prognosis in EMPD. Cases with BRCA2-mutant cancer might also have a worse prognosis in EMPD. In addition, CGP testing before PS deteriorates is preferable, considering the observed median survival of individuals undergoing CGP tests in an ECOG-PS-1 condition was less than 9 months.

Introduction

Extramammary Paget’s disease (EMPD) is a rare cutaneous malignancy primarily found in regions rich in apocrine glands, particularly the genital area1,2. While surgical excision remains the standard treatment, systemic therapy is required for advanced cases1,3. From a pathogenetic perspective, previous studies have reported that HER2 overexpression and the PI3K/AKT/mTOR pathway are associated with the pathogenesis in a subset of EMPD, and genetic variants in receptor tyrosine kinases (RTKs) including ERBB2, ERBB3 and FGFR1, and the PI3K/AKT/mTOR-pathway-associated genes are frequently observed1,4-6.

Since June 2019 in Japan, comprehensive genomic profiling (CGP) tests have been covered by public health insurance7. Our previous work reported the mutational landscape of EMPD using the Center for Cancer Genomics and Advanced Therapeutics (C-CAT) database, which compiled CGP results with clinical information in Japan. In the study, oncogenic variants in RTKs and PI3K/AKT/mTOR-pathway-associated genes were frequently observed, supporting the involvement of the pathway in EMPD8. However, due to limited data accumulated in the database, our previous study did not allow us to perform prognostic analysis.

In relatively rare cancers like EMPD, databases containing genetic variants and prognostic data are invaluable. However, when analyzing such real-world clinicogenomic data, it is crucial to carefully consider potential biases. Individuals who have not undergone CGP are not included in the database, leading to selection biases. Moreover, CGP tests are often submitted after the application of initial chemotherapy, providing a time gap between the survival time to be analyzed and the practical observation time in the database, known as left truncation9,10. Unless the length bias caused by left truncation is corrected, overall survival would be overestimated9,10. Specifically, overall survival differed by more than 12 months in patients with stage IV colorectal or non-small-cell lung cancer depending on whether the length bias was adjusted9. Furthermore, the correlation between left truncation and overall survival should be examined when making this adjustment because a Kaplan-Meier estimator, an established adjustment method, assumes that this correlation does not exist9,10.

In addition to alterations in RTKs- and PI3K/AKT/mTOR-pathway-associated genes as described above, CDKN2A loss is frequently detected in EMPD. A previous study encompassing early stages of EMPD reported CDKN2A variants, predominantly characterized by loss, at a rate of 34%4. In our preceding investigation targeting on metastatic EMPD, CDKN2A variants, predominantly exhibiting loss, were observed at a higher frequency (68%)8. CDKN2A is a tumor suppressor gene linked to the cell cycle, and the association between CDKN2A loss and tumor progression or poor prognosis has been reported in other cancers11-17. However, the involvement of CDKN2A loss in EMPD remains unexplored.

Here we analyzed the C-CAT database in EMPD to examine the impact of common genetic alterations including CDKN2A variants on prognosis.

Materials and methods

Study design

This is a retrospective cohort study using the C-CAT database, a data center established by The Ministry of Health, Labour and Welfare of Japan. This database contains clinical data and CGP results of individuals who underwent CGP in Japan. Prognostic information following CGP test submission is updated in the database by the attending physicians. While specific data on race and ethnicity were unavailable, most individuals represented the Japanese population. Cases with EMPD registered in the C-CAT database until January 29, 2024, were included. For CGP tests, FoundationOne® CDx(F1CDx) and FoundationOne® Liquid CDx (F1LiquidCDx) can detect 324 genetic alterations, while the OncoGuideTM NCC Oncopanel System (NCC Oncopanel) can analyze 124. The top 20 genes with high variant frequencies were subjected to the survival analysis. This study was approved by the research ethics committee of the Faculty of Medicine of The University of Tokyo (approval number: 2021341G). Research use of the database was also approved by the C-CAT Data Utilization Review Board (approval number. CDU2022-026N). Written informed consent for the secondary use of data from the C-CAT database was obtained from each individual at the time of ordering CGP testing at each registered facility.

Survival analysis

The timing of CGP test submission depends on individual cases, and the stage at the time of diagnosis was missing in the database. Therefore, survival analyses were conducted using the date of chemotherapy initiation as the reference point. Considering CGP tests for EMPD cases can be submitted after the point at which chemotherapy is required, the CGP registration date was used as a substitute for the chemotherapy initiation date for individuals who underwent CGP testing before starting chemotherapy and lacked documented chemotherapy start dates. Cases in which chemotherapy was initiated prior to registration for CGP, but lacked documentation of the chemotherapy start date, were excluded from the analysis. Cases whose first chemotherapy was recorded as neoadjuvant or adjuvant and those without metastasis or lacking metastatic data were also excluded from the prognostic analysis. These exclusions resulted in our cohort enriched for individuals who received chemotherapy for metastatic disease.

In addition to the primary analysis of survival from the chemotherapy initiation, we also conducted a secondary analysis examining the survival from CGP registration based on Eastern Cooperative Oncology Group performance status (ECOG-PS) at the time of CGP enrollment.

Statistical method of survival analysis

All statistical analyses were performed using R version 4.3.2. The calculation of Kendall's tau statistic calculation was performed using the tranSurv package as reported previously18. Survival analysis involved the use of various techniques such as the Kaplan-Meier estimator, log-rank test, and Cox regression analysis, all facilitated by the survival package19. Hazard ratios (HR) were calculated using Cox regression analysis.

Results

Patient characteristics

A total of 167 cases with EMPD were registered in the C-CAT database. Out of the total 167 cases, prognostic data for survival from chemotherapy initiation were available in 111 cases. Among them, 5 cases were excluded due to neoadjuvant or adjuvant chemotherapy being recorded as their initial treatment. Additionally, 24 cases were included for whom the start date of chemotherapy was unavailable but CGP test was submitted before starting chemotherapy. In these cases, the CGP registration date was utilized as an alternative to the chemotherapy initiation date. Finally, three cases without metastasis or lacking metastatic data were excluded. Consequently, 127 cases of metastatic EMPD were subjected to prognostic analysis.

Patient characteristics are summarized in Table 1. The median age of the cohort was 71 years (range, 47-90), with 95 cases (75%) being male. Most individuals (97%) had a good ECOG-PS (0 or 1), and all the cases were metastatic disease. The median follow-up time after CGP enrollment was 237 days (IQR, 116-444 days; range, 13-1381 days). Chemotherapy was started before CGP enrollment in 94 cases, while CGP test was submitted before starting chemotherapy in 33 cases. Among 33 individuals who undergone CGP before chemotherapy initiation, chemotherapy start dates were available in 9 cases. The duration between the chemotherapy initiation and CGP registration exhibited considerable variability among individuals; however, in cases where CGP submission preceded chemotherapy initiation, chemotherapy was promptly commenced (Table 2). Specifically, the median time from chemotherapy initiation to CGP submission was 248.5 days (IQR, 69.75-683.75 days; range, 1-1879 days), while the median duration from CGP submission to chemotherapy initiation was 11 days (IQR, 5-12 days; range, 1-23 days).

Prognostic analysis

The median survival times from chemotherapy initiation and CGP registration were 865 days (95% CI, 676-1186 days; Figure 1a) and 395 days (95% CI, 308-621 days; Figure 1b), respectively. ECOG-PS data were exclusively available at the time of CGP submission. While most cases underwent CGP testing with a favorable ECOG-PS of 0 or 1, individuals with an ECOG-PS 1 at the time of CGP enrollment experienced significantly shorter survival time compared to those with an ECOG-PS 0 (Figure 1c, log-rank test, p=0.034; median survival time, 531 vs. 259 days). Before conducting the analysis of survival from chemotherapy initiation, we evaluated the length bias caused by left truncation, the time gap from the start of chemotherapy to CGP enrollment. The conditional Kendall’s tau statistic was calculated for the 98 cases for whom prognostic data were available and CGP tests were submitted after the initiation of chemotherapy. The statistic was close to zero and not significant (-0.072, p=0.33), indicating no correlation between left truncation and overall survival. Therefore, the Kaplan-Meier estimator, the established method was applied to adjust for the length bias. The adjusted survival curve yielded a median survival of 552 days (95% CI, 404-676 days; Figure 1d).

Gene variants and prognosis

The mutational landscape showed variants at a high frequency in genes such as CDKN2A, ERBB2, CDKN2B, MTAP, TP53, ERBB3, and PIK3CA (Figure 2a). Hazard ratios (HR) and p-values were calculated using Cox regression for the top 20 genes with high variant frequencies. The correlations between gene variants and prognosis were evaluated primarily in the length-bias-adjusted model, and also in the unadjusted model (Figure 2b and Figure S1, respectively). In both models, the significant correlation with a poorer prognosis was identified in BRCA2-mutant group (length-bias-adjusted model, HR=2.97, 95% CI 1.46-6.01, p=0.003; unadjusted model, HR=2.39, 95% CI 1.20-4.77, p=0.014) and CDKN2A-mutant group (length-bias-adjusted model, HR=1.81, 95% CI 1.06-3.07 p=0.029; unadjusted model, HR=1.73, 95% CI 1.03-2.93, p=0.039). Furthermore, even when limiting the analysis to BRCA2 wild-type cases, a statistically significant difference was still observed between

9 the CDKN2A wild-type and CDKN2A variant groups (length-bias-adjusted model, HR=2.01, 95% CI 1.12-3.62, p=0.019; unadjusted model, HR=1.93, 95% CI 1.08-3.45, p=0.026; Figure S2). Similarly, limiting the analysis to CDKN2A wild-type cases, a statistically significant difference was again found between the BRCA2 wild-type and BRCA2 mutant groups (length-bias-adjusted model, HR=4.12, 95% CI 1.46-11.59, p=0.007; unadjusted model, HR=2.85, 95% CI 1.04-7.81, p=0.042; Figure S2).

BRCA2 variants and prognosis

Cases with BRCA2 variants exhibited the most significantly poor prognosis. In the 22 cases with BRCA2 variants, prognostic data were available for 18 individuals. The most frequently observed variants were missense variants classified as variants of uncertain significance (VUS, Figure 2a, Figure 114). Only two variants were identified as pathogenic or likely pathogenic: allelic loss (pathogenic) and frameshift variant (E2193fs*11, likely pathogenic). FoundationOne® CDx, the CGP test used in most cases, could not identify whether variants were germline, while OncoGuideTM NCC Oncopanel System (NCC Oncopanel) could. Then, germline variant was detected by NCC Oncopanel in one case with K1132R and V2109I co-variant: VUS and likely benign, respectively.

CDKN2A variants and prognosis

Among 167 cases, CDKN2A variants were the most frequently observed (n=93, 56%). The identified CDKN2A variants were loss-of-function (LoF) pathogenic variants in all except for three VUS. In the CDKN2A variants, loss was the most prevalent finding (n=80, 86%), while nonsense or frameshift variants were observed in 6 cases, and missense variants in 9 cases (Figure 2a, Figure 4).

In the survival analysis, the CDKN2A-mutant group had a poorer prognosis than the CDKN2A-wild-type group. Furthermore, we obtained similar observations when the analysis in the length-bias-adjusted model was confined to the CDKN2A-loss cases versus the CDKN2A-wild-type cases (HR=1.77, 95% CI 1.03-3.04, p=0.039).

Discussion

This study analyzed the correlation between common genetic alterations and prognosis in EMPD using the C-CAT database. Our prognostic analysis indicated a significant correlation of CDKN2A variants, predominantly LoF, and BRCA2 variants with poor prognosis.

In the analysis of survival from CGP enrollment, individuals with ECOG-PS 1 at the time of CGP enrollment had significantly poorer prognosis than those with ECOG-PS 0. The observed median survival of cases undergoing CGP tests in an ECOG-PS-1 condition was less than 9 months (Figure 1c). Even if actionable genetic alterations are identified, the individual's condition may deteriorate further, making it challenging to administer recommended drugs or enroll in clinical trials. Literature from other cancers suggests that early timing of CGP testing may be clinically beneficial in guiding subsequent anticancer therapies20,21. Our findings suggest that CGP testing before PS deteriorates may lead to better outcomes.

Given the variation in the timing of CGP testing among individual cases, the duration of survival from chemotherapy initiation holds greater clinical significance than that from CGP enrollment in practical terms. In this study, the median survival time from the start of chemotherapy was 865 days (Figure 1a). Although the prognosis following chemotherapy initiation varies among reports due to the rarity of the disease and the lack of established treatment, the prognosis observed in this study was longer than previously reported22-26. This discrepancy can be attributed to the length bias caused by left truncation inherent in the database9,10, emphasizing the need for appropriate adjustment to obtain accurate prognosis estimates. Adjustment for this bias is feasible using the Kaplan-Meier estimator only under the condition that left truncation time and overall survival are not correlated. In our cohort, the conditional Kendall’s tau statistic demonstrated proximity to zero, indicating no correlation between left truncation time and overall survival. Consequently, successful adjustment was achieved (Figure 1d). A previous report indicated that prognosis varies depending on the presence or absence of internal organ metastasis even among Stage IV EMPD cases27. Given the potential differences in such backgrounds, direct comparisons between cohorts may not be feasible; therefore, while it serves as a reference, the adjusted survival curves resemble those reported for the largest historical data for Stage IV EMPD28.

In the analysis of the correlation between commonly altered genes and EMPD prognosis, cases with BRCA2 variants exhibited significantly poorer prognosis than those with BRCA2 wild-type (Figure 2b, 3a, 3b). BRCA2 variants were detected in 13% of the cases (n=22), most of which were missense VUS (Figure 2a, Figure 4). The majority of BRCA2 missense variants are currently categorized as VUS; therefore, the accurate classification of missense variants in BRCA2 as pathogenic holds significant importance29,30. BRCA2 plays a role in DNA double-strand break repair, and its germline pathogenic variants are associated with increased risk of breast and ovarian cancers31. In breast cancer, BRCA2 variants are associated with estrogen receptor expression32. Despite limited literature on BRCA2 alterations in EMPD6, given that estrogen receptor expression is observed in a subset of EMPD33, BRCA2 variants might potentially contribute to EMPD tumorigenesis. However, the significant difference of survival in BRCA2-mutant cases should be interpreted cautiously. The Kaplan-Meier estimator is sensitive to the impact of a small number of deaths in dealing with a small subset, especially when left truncation is present. Considering the limited number of BRCA2-mutant cases and the predominance of VUS, it is desirable to accumulate more cases to validate our results. Despite these limitations, our findings offer valuable insights suggesting a prognostic impact of BRCA2 variants on EMPD.

The CDKN2A-mutant group had a significantly poorer prognosis than the CDKN2A-wild-type

8 group (Figure 2b, 3c, 3d). CDKN2A was the most frequently altered gene in our cohort (56%). Although F1CDx, the most commonly used CGP test in our cohort, could not identify whether detected variants were somatic or germline, considering the fact that germline CDKN2A variants are quite rare in Japan34, the identified CDKN2A variants were almost certainly somatic. The identified CDKN2A variants were predominantly loss, and the correlation between CDKN2A loss and poor prognosis has been reported in other cancers16,17,35. Furthermore, a previous study investigating genetic variants in 87 EMPD cases, including early stages (detailed stage data were not provided), reported CDKN2A

15 loss in 32% of cases4. The higher rate of CDKN2A loss in this study (48%) may be due to the greater proportion of metastatic cases. As reported in some types of cancer, CDKN2A loss may also be associated with tumor progression and poor prognosis in EMPD.

This study has several limitations. First, this is a retrospective cohort study utilizing the limited database called C-CAT. Our investigation could not encompass cases who did not undergo CGP testing or genes not covered by CGP. The absence of certain information, such as clinical stage and performance status at the time of chemotherapy initiation, compromises the precision of our analysis. It is desirable that the results obtained from this study should be validated using other methods, preferably through a prospective approach. Secondly, the cases registered in the database are predominantly Japanese, which may limit the generalizability of our findings to EMPD cases of other races.

In summary, our study explored genetic variants and their impact on prognosis in EMPD using the C-CAT database. Prognostic analysis using this kind of database requires careful consideration of biases, particularly those introduced by left truncation. In our study, appropriate adjustments allowed us to derive survival curves approximating historical data. Our survival analysis provides valuable insights into the correlation between common genetic alterations and prognosis in EMPD, a topic that has been scarcely reported. Our findings propose BRCA2 and CDKN2A variants as novel poor prognostic factors. Despite several limitations, given the rarity of this disease, we believe this study constitutes a valuable contribution to effective use of CGP testing and deeper comprehension of EMPD tumorigenesis.

References

Fukuda K, Funakoshi T. Metastatic Extramammary Paget's Disease: Pathogenesis and Novel Therapeutic Approach. Front Oncol 2018; 8: 38.

Morris CR, Hurst EA. Extramammary Paget Disease: A Review of the Literature-Part I: History, Epidemiology, Pathogenesis, Presentation, Histopathology, and Diagnostic Work-up. Dermatol Surg 2020; 46: 151-8.

Morris CR, Hurst EA. Extramammary Paget's Disease: A Review of the Literature Part II: Treatment and Prognosis. Dermatol Surg 2020; 46: 305-11.

Ishida Y, Kakiuchi N, Yoshida K et al. Unbiased Detection of Driver Mutations in Extramammary Paget Disease. Clin Cancer Res 2021; 27: 1756-65.

Stasenko M, Jayakumaran G, Cowan R et al. Genomic Alterations as Potential Therapeut i c Targets in Extramammary Paget's Disease of the Vulva. JCO Precis Oncol 2020; 4.

Kiniwa Y, Yasuda J, Saito S et al. Identification of genetic alterations in extramammary Paget disease using whole exome analysis. J Dermatol Sci 2019; 94: 229-35.

Ebi H, Bando H. Precision Oncology and the Universal Health Coverage System in Japan. JCO Precis Oncol 2019; 3.

Iwasawa O, Miyagawa T, Omatsu J et al. Mutational landscape of advanced extramammary Paget disease with comprehensive genomic profiling tests: A cohort study in a Japanese real-world setting. J Am Acad Dermatol 2023.

Brown S, Lavery JA, Shen R et al. Implications of Selection Bias Due to Delayed Study Entry in Clinical Genomic Studies. JAMA Oncol 2022; 8: 287-91.

Tamura T, Ikegami M, Kanemasa Y et al. Selection bias due to delayed comprehensive genomic profiling in Japan. Cancer Sci 2023; 114: 1015-25.

Qiao J, Tian Y, Cheng X et al. Deletion Leading to Hematogenous Metastasis of Human Gastric Carcinoma. Front Oncol 2021; 11: 801219.

Zeng H, Jorapur A, Shain AH et al. Bi-allelic Loss of CDKN2A Initiates Melanoma Invasion via BRN2 Activation. Cancer Cell 2018; 34: 56-68.e9.

Wach J, Basaran AE, Arlt F et al. CDKN2A/B deletions are strongly associated with meningioma progression: a meta-analysis of individual patient data. Acta Neuropathol Commun 2023; 11: 189.

Zhang W, Kuang P, Liu T. Prognostic significance of CDKN2A/B deletions in acute lymphoblastic leukaemia: a meta-analysis. Ann Med 2019; 51: 28-40.

Vij M, Cho BB, Yokoda RT et al. P16 immunohistochemistry is a sensitive and specific surrogate marker for CDKN2A homozygous deletion in gliomas. Acta Neuropathol Commun 2023; 11: 73.

Reis GF, Pekmezci M, Hansen HM et al. CDKN2A loss is associated with shortened overall survival in lower-grade (World Health Organization Grades II-III) astrocytomas. J Neuropathol Exp Neurol 2015; 74: 442-52.

Bortolotto S, Chiadò-Piat L, Cavalla P et al. CDKN2A/p16 inactivation in the prognosis of oligodendrogliomas. Int J Cancer 2000; 88: 554-7.

Chiou SH, Austin MD, Qian J et al. Transformation model estimation of survival under dependent truncation and independent censoring. Stat Methods Med Res 2019; 28: 3785- 98.

Therneau T. A package for survival analysis in R. In, Vol. 2024. 2024.

Thapa B, Ahmed G, Szabo A et al. Comprehensive genomic profiling: Does timing matter? Front Oncol 2023; 13: 1025367.

Matsubara J, Mukai K, Kondo T et al. First-Line Genomic Profiling in Previously Untreated Advanced Solid Tumors for Identification of Targeted Therapy Opportunities. JAMA Netw Open 2023; 6: e2323336.

Matsushita S, Fujii K, Kajihara I et al. Efficacy of S-1 plus docetaxel in the treatment of metastatic extramammary Paget's disease: a multicentre retrospective study. Br J Dermatol 2021; 185: 458-60.

Yoshino K, Fujisawa Y, Kiyohara Y et al. Usefulness of docetaxel as first -line chemotherapy for metastatic extramammary Paget's disease. J Dermatol 2016; 43: 633-7.

Tokuda Y, Arakura F, Uhara H. Combination chemotherapy of low-dose 5-fluorouracil and cisplatin for advanced extramammary Paget's disease. Int J Clin Oncol 2015; 20: 194- 7.

Kato M, Yoshino K, Maeda T et al. Single-agent taxane is useful in palliative chemotherapy for advanced extramammary Paget disease: a case series. Br J Dermatol 2019; 181: 831-2.

Oashi K, Tsutsumida A, Namikawa K et al. Combination chemotherapy for metastatic extramammary Paget disease. Br J Dermatol 2014; 170: 1354-7.

Funakoshi T, Nakamura Y, Hirai I et al. Proposal for the optimal staging of extramammary Paget's disease: A retrospective analysis of 57 cases with distant metastasis. Exp Dermatol 2024; 3 3 : e15017.

Ohara K, Fujisawa Y, Yoshino K et al. A proposal for a TNM staging system for extramammary Paget disease: Retrospective analysis of 301 patients with invasive primary tumors. J Dermatol Sci 2016; 83: 234-9.

Sessa G, Ehlén Å, von Nicolai C et al. Missense Variants of Uncertain Significance: A Powerful Genetic Tool for Function Discovery with Clinical Implications. Cancers (Basel) 2021; 1 3 .

Hart SN, Hoskin T, Shimelis H et al. Comprehensive annotation of BRCA1 and BRCA2 missense variants by functionally validated sequence-based computational prediction models. Genet Med 2019; 21: 71-80.

Kuchenbaecker KB, Hopper JL, Barnes DR et al. Risks of Breast, Ovarian, and Contralateral Breast Cancer for BRCA1 and BRCA2 Mutation Carriers. JAMA 2017; 317: 2402-16.

Pujol P, Yauy K, Coffy A et al. Predominance of. Cancers (Basel) 2022; 14.

Zhou S, Zhong W, Mai R et al. Mammary and Extramammary Paget's Disease Presented Different Expression Pattern of Steroid Hormone Receptors. Biomed Res Int 2017; 2017: 3768247.

Kiyozumi Y, Matsubayashi H, Todaka A et al. Two Japanese families with familial pancreatic cancer with suspected pathogenic variants of CDKN2A: a case report. Hered Cancer Clin Pract 2024; 22: 11.

Xue L, Tang W, Zhou J et al. Next-generation sequencing identifies CDKN2A alterations as prognostic biomarkers in recurrent or metastatic head and neck squamous cell carcinoma predominantly receiving immune checkpoint inhibitors. Front Oncol 2023; 13: 1276009.

Authors: Okuto Iwasawa, Masachika Ikegami, Takuya Miyagawa, Hiromichi Morita, Hinako Saito,1 Issei Omori, Kentaro Awaji, Jun Omatsu, Daisuke Yamada, Hidenori Kage, Katsutoshi Oda, Shinichi Sato1 and Hayakazu Sumida1

Ethics statement: This study was approved by the research ethics committee of the Faculty of Medicine of the University of Tokyo (approval number: 2021341G). Research use of the database was also approved by the C-CAT Data Utilization Review Board (approval number. CDU2022-026N).

Patient consent: Written patient consent for publication was obtained.

Funding sources: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflicts of interest: The authors have no conflict of interest.