ABSTRACT
Objectives
The aim of this study was to determine whether pretreatment PD‐L1 CPS is prognostically associated with survival in patients with HPSCC treated nonsurgically and without immunotherapy, and to assess overall survival (OS) in actively treated patients without distant metastases according to CPS < 1 versus CPS ≥ 1.
Methods
This retrospective single‐center study included 71 patients; the primary cohort comprised 56 M0 patients receiving active treatment. CPS was categorized as < 1 versus ≥ 1. Overall survival (OS) and progression‐free survival (PFS), defined as progression or death, were analyzed using Kaplan–Meier and adjusted Cox models for age, ECOG performance status, grouped T stage, and nodal status. Sensitivity analyses examined broader cohorts and treatment covariates.
Results
In the primary cohort, 47 deaths and 49 PFS events occurred. Median OS was 9 versus 18 months for CPS < 1 versus ≥ 1 (p = 0.001). For CPS ≥ 1 versus < 1, the adjusted hazard ratio (HR) was 0.243 (95% CI, 0.100–0.591; p = 0.002). Median PFS was 5 versus 12 months (p = 0.004), with an adjusted HR of 0.356 (95% CI, 0.149–0.847; p = 0.020). Sensitivity analyses showed consistent OS estimates (adjusted HRs, 0.292–0.455), whereas PFS estimates were attenuated (adjusted HRs, 0.431–0.603; p = 0.051–0.212). Continuous CPS and CPS ≥ 20 were not associated with either outcome; three‐category analyses suggested that poorer outcomes in CPS < 1 drove the association.
Conclusions
CPS ≥ 1 was associated with longer OS and PFS in the primary cohort, with consistent OS estimates across sensitivity analyses. These hypothesis‐generating findings require external validation and do not support CPS as a standalone treatment‐selection marker.
Level of Evidence
3.
Keywords: combined positive score, hypopharyngeal squamous cell carcinoma, overall survival, PD‐L1, progression‐free survival
1. Introduction
Hypopharyngeal squamous cell carcinoma (HPSCC) is among the most aggressive head and neck squamous cell carcinoma (HNSCC) subsites. It is often diagnosed at an advanced stage because early symptoms are nonspecific and hypopharyngeal anatomy facilitates submucosal extension and early lymphatic spread [1, 2]. Many patients therefore present with locally advanced disease, cervical nodal metastases, impaired performance status, and high risks of locoregional failure and distant metastasis. Despite advances in diagnosis, radiotherapy, systemic therapy, and multidisciplinary care, long‐term survival remains poor [3, 4].
Management depends on tumor stage, resectability, laryngeal function, comorbidities, performance status, and expected functional outcomes. Nonsurgical management may reflect comorbidities, poor general condition, refusal of surgery, organ‐preservation intent, functional considerations, or palliative context. Definitive radiotherapy, concurrent chemoradiotherapy, and induction chemotherapy followed by radiotherapy or chemoradiotherapy remain important strategies [5, 6]. Comparisons with surgery are limited by differences in patient selection, stage, treatment protocols, and outcome definitions [7, 8, 9], while larynx‐preservation and definitive chemoradiotherapy studies continue to show limited survival [10, 11].
The programmed death‐1/programmed death‐ligand 1 (PD‐1/PD‐L1) pathway is central to recurrent and metastatic HNSCC. Immune checkpoint inhibitors improved outcomes in CheckMate 141 and KEYNOTE‐048 [12, 13], increasing the clinical relevance of PD‐L1 assessment. The combined positive score (CPS) incorporates PD‐L1‐positive tumor cells, lymphocytes, and macrophages relative to the total number of viable tumor cells.
However, the prognostic value of PD‐L1 CPS without immunotherapy remains uncertain. Antibody clones, staining platforms, tissue sources, scoring systems, cutoffs, and interpretative criteria may influence classification [14, 15, 16]. CPS may reflect the tumor immune microenvironment more broadly than tumor‐cell‐only scoring but remains affected by intratumoral heterogeneity, limited biopsy material, interobserver variability, and scoring accuracy in small samples [15, 17]. Standardized interpretation and careful methodological reporting are therefore essential [16, 18]. HPSCC‐specific evidence remains limited and inconsistent.
The aim of this study was to determine whether pretreatment PD‐L1 CPS is prognostically associated with survival in patients with HPSCC treated nonsurgically and without immunotherapy, and to assess overall survival (OS) in actively treated patients without distant metastases according to CPS < 1 versus CPS ≥ 1.
2. Materials and Methods
2.1. Study Design and Patient Cohort
This retrospective cohort study included all consecutive adults with newly diagnosed, histopathologically confirmed hypopharyngeal squamous cell carcinoma whose initial multidisciplinary treatment decision was nonsurgical management at the University Clinical Center from January 1, 2014, to December 31, 2018. Eligibility was determined from initial diagnostic and treatment‐planning records, independently of subsequent treatment completion or outcomes. Additional requirements were no immunotherapy exposure and availability of a pretreatment primary‐tumor biopsy obtained and processed at our institution, with an archival formalin‐fixed, paraffin‐embedded block suitable for PD‐L1 assessment. Exclusion criteria were diagnosis before the study period, surgical treatment, nonsquamous histology, biopsy performed elsewhere, synchronous primary malignancy, or unavailable archival tissue.
No formal sample size calculation was performed. Data were extracted from electronic medical records and institutional documentation. Smoking and alcohol histories were available for 64 and 60 patients, respectively. Comorbidity was recorded as present or absent because type and severity were inconsistently documented; p16/HPV status was unavailable because testing was not routine.
TNM categories and stage groups were assigned according to the eighth AJCC/UICC edition, with retrospective reclassification when required, and ECOG performance status was recorded on a 0–4 scale. Vital status and dates of death were verified through the national death registry. The study complied with the Declaration of Helsinki and was approved by the Institutional Ethics Committee of the University Clinical Center (No. 1322/IV‐10; April 18, 2022), which waived study‐specific informed consent because of the retrospective design and because most patients were deceased at the time of ethical review.
2.2. Treatment Modalities and Response Assessment
All patients were managed nonsurgically without immunotherapy. Multidisciplinary decisions considered disease extent, performance status, comorbidities, anesthesiological risk, and patient preference. Pretreatment intent was classified as curative, palliative, or supportive care only. Curative treatment included definitive radiotherapy (RT) or chemoradiotherapy (CRT), or induction chemotherapy followed by definitive local treatment; palliative treatment included RT or chemotherapy for symptom control, whereas supportive care involved no active anticancer treatment. In potentially resectable disease, surgery was omitted because of patient refusal, contraindicating comorbidities, or high anesthesiological risk, or larynx‐preservation intent. Treatment intent was curative in 57 patients, palliative in 4, and supportive in 10.
Initial modality, which did not necessarily represent the complete treatment sequence, was chemotherapy in 42 patients, RT in 6, CRT in 13, and supportive care in 10. RT used 3D‐CRT, IMRT, or VMAT; definitive treatment was generally 70 Gy in 35 fractions, and palliative schedules were 40 Gy in 20 fractions or 31 Gy in 15 fractions. Concurrent CRT used weekly cisplatin 40 mg/m2. Of 42 patients receiving chemotherapy, 40 received induction treatment, usually cisplatin plus 5‐fluorouracil every 3 weeks for three planned cycles before definitive RT or CRT; 2 received palliative chemotherapy. Sixty of 61 patients initiating active treatment completed it; one died from cardiac complications after the first induction cycle.
Response was assessed by imaging approximately 6 weeks after treatment completion or according to institutional protocols using RECIST version 1.1. Objective response comprised complete or partial response, and disease control comprised complete response, partial response, or stable disease. Ten supportive‐care patients and the patient who died before formal assessment were not evaluable.
2.3. Immunohistochemical Assessment of PD‐L1 CPS
PD‐L1 CPS was assessed on pretreatment formalin‐fixed, paraffin‐embedded biopsy specimens from primary hypopharyngeal tumors; no nodal or distant metastatic specimens were used. When multiple blocks were available, the most representative block with the greatest amount of viable tumor was selected. All specimens were fixed in 10% neutral‐buffered formalin and processed at the same institution under a uniform protocol.
Staining was performed using the commercially validated PD‐L1 IHC 22C3 pharmDx assay (Dako/Agilent) according to the manufacturer's instructions, with positive and negative controls included in each run. A minimum of 100 viable tumor cells was required. Technically inadequate staining or fixation‐related artifacts prompted repeat staining on an additional section. All 71 cases met the adequacy criteria, and none was excluded.
Two pathologists independently assessed PD‐L1 expression while blinded to clinical outcomes; discrepancies were resolved by joint review and consensus. CPS was calculated as the number of PD‐L1‐positive tumor cells, lymphocytes, and macrophages divided by the total number of viable tumor cells and multiplied by 100, with values capped at 100. For the primary analyses, the predefined cutoff was CPS < 1 versus CPS ≥ 1.
2.4. Outcome Definitions
The first disease‐related admission to our institution for evaluation and management of subsequently confirmed HPSCC was used as the common time origin for overall survival (OS) and progression‐free survival (PFS). OS was defined from this date to death from any cause; patients alive at the administrative end of follow‐up on December 15, 2022, were censored on that date. PFS was defined from the same date to first documented radiologic or clinical progression or death, whichever occurred first. Patients alive without progression were censored at their last available disease assessment.
2.5. Statistical Analysis
Statistical analyses were performed using IBM SPSS Statistics v25.0. All tests were two‐sided, with p < 0.05 considered significant; no multiplicity adjustment was applied to exploratory or sensitivity analyses. Categorical variables were summarized as n (%) and continuous variables as median (interquartile range). CPS groups (< 1 vs. ≥ 1) were compared using the Mann–Whitney U, Pearson chi‐square, or Fisher's exact test, as appropriate.
The primary analytic cohort comprised M0 patients receiving active anticancer treatment; sensitivity analyses included the active‐treatment, M0, and entire cohorts. OS and PFS were estimated by Kaplan–Meier analysis and compared using the log‐rank test; median survival and reverse Kaplan–Meier follow‐up were reported with 95% confidence intervals.
CPS associations with OS and PFS were examined using univariable and multivariable Cox regression. Primary models included CPS, age, ECOG performance status, grouped T stage, and nodal status. Comorbidity and treatment intent were added separately to limit overparameterization, and active‐treatment status was added to selected M0 and entire‐cohort models. Proportional hazards and age linearity were assessed using covariate‐by‐log‐time interactions and a centered quadratic age term, respectively.
Exploratory analyses assessed CPS continuously, at 20, and in three categories (< 1, 1–19, and ≥ 20), including direct comparison of ≥ 20 versus 1–19. ORR, DCR, and best‐response distributions were compared using exact tests. The patient who died before formal response assessment was classified as a nonresponder for ORR/DCR but not as having progressive disease.
Missing data were not imputed; smoking and alcohol analyses used available cases, and Cox models used complete cases. Disease‐specific survival was not analyzed because reliable cause‐of‐death data were unavailable. Reporting followed STROBE.
3. Results
3.1. Patient Selection and Analytic Cohorts
Figure 1 summarizes patient identification and selection. Of 146 consecutive patients with hypopharyngeal carcinoma identified in institutional records, 71 met the eligibility criteria and formed the final nonsurgical, nonimmunotherapy HPSCC cohort. Active anticancer treatment was initiated in 61 patients, whereas 10 received supportive care only. The primary analytic cohort comprised 56 M0 patients receiving active treatment (7 with CPS < 1 and 49 with CPS ≥ 1). Sensitivity cohorts included the 61 actively treated patients, 64 M0 patients, and the full cohort of 71.
FIGURE 1.
Flow diagram of patient identification, eligibility assessment, exclusions, and derivation of the study and analytic cohorts.
3.2. PD‐L1 CPS Expression and Baseline Characteristics
The median PD‐L1 CPS in the entire cohort was 6 (IQR, 21; range, 0–98); 13 patients (18.3%) had CPS < 1 and 58 (81.7%) had CPS ≥ 1. Initial assessments were concordant in 64 of 71 cases (90.1%), whereas 7 cases were resolved by consensus review.
The median age was 61 years, 62 patients (87.3%) were male, and the piriform sinus was the most common primary site (76.1%). ECOG performance status was 0–1 in 55 patients (77.5%). T4a–T4b disease was present in 59.2%, nodal involvement in 78.9%, and distant metastases in 9.9%.
In the entire cohort, distant metastases (30.8% vs. 5.2%; p = 0.018), stage distribution (p = 0.044), and documented comorbidity (92.3% vs. 62.1%; p = 0.048) differed between CPS groups; no other evaluated baseline or treatment characteristics differed significantly (Supplemental Table 1). No significant baseline or treatment differences were observed within the primary analytic cohort (Table 1).
TABLE 1.
Baseline and treatment characteristics of the primary analytic cohort according to PD‐L1 combined positive score status.
| Characteristics | Total (n = 56) | CPS < 1 (n = 7) | CPS ≥ 1 (n = 49) | p |
|---|---|---|---|---|
| Age, years, median (IQR) | 61 (11) | 64 (12) | 61 (11) | 0.363 |
| Sex | 0.577 | |||
| Male | 48 (85.7) | 7 (100.0) | 41 (83.7) | |
| Female | 8 (14.3) | 0 (0.0) | 8 (16.3) | |
| ECOG performance status | 0.501 | |||
| ECOG 0–1 | 51 (91.1) | 6 (85.7) | 45 (91.8) | |
| ECOG ≥ 2 | 5 (8.9) | 1 (14.3) | 4 (8.2) | |
| Grouped T stage | 1.000 | |||
| T1–T3 | 26 (46.4) | 3 (42.9) | 23 (46.9) | |
| T4a–T4b | 30 (53.6) | 4 (57.1) | 26 (53.1) | |
| Nodal status | 0.635 | |||
| N0 | 12 (21.4) | 2 (28.6) | 10 (20.4) | |
| N1–N3 | 44 (78.6) | 5 (71.4) | 39 (79.6) | |
| Smoking status a | 0.709 | |||
| Never smoker | 4 (7.8) | 0 (0.0) | 4 (8.9) | |
| Former smoker | 5 (9.8) | 1 (16.7) | 4 (8.9) | |
| Current smoker | 42 (82.4) | 5 (83.3) | 37 (82.2) | |
| Missing, n | 5 | 1 | 4 | |
| Alcohol consumption b | 0.685 | |||
| No reported consumption | 23 (42.6) | 2 (28.6) | 21 (44.7) | |
| Any reported consumption | 31 (57.4) | 5 (71.4) | 26 (55.3) | |
| Missing, n | 2 | 0 | 2 | |
| Documented comorbidity c | 0.082 | |||
| None | 19 (33.9) | 0 (0.0) | 19 (38.8) | |
| At least one | 37 (66.1) | 7 (100.0) | 30 (61.2) | |
| Initial treatment modality d | 1.000 | |||
| Chemotherapy | 37 (66.1) | 5 (71.4) | 32 (65.3) | |
| Radiotherapy | 6 (10.7) | 1 (14.3) | 5 (10.2) | |
| Chemoradiotherapy | 13 (23.2) | 1 (14.3) | 12 (24.5) | |
| Treatment intent | 0.423 | |||
| Curative | 52 (92.9) | 6 (85.7) | 46 (93.9) | |
| Palliative | 4 (7.1) | 1 (14.3) | 3 (6.1) |
Note: Data are presented as n (%) unless otherwise indicated. Age is presented as median (IQR). The primary analytic cohort comprised 56 patients without distant metastases who received active anticancer treatment. Age was compared using the exact Mann–Whitney U test, and categorical variables were compared using Fisher's exact or Fisher–Freeman–Halton exact tests, as appropriate. All tests were two‐sided. Percentages and p‐values for smoking status and alcohol consumption were calculated using available data; missing values were not imputed.
Abbreviations: CPS, combined positive score; ECOG, Eastern Cooperative Oncology Group; IQR, interquartile range; N, regional lymph node category; PD‐L1, programmed death‐ligand 1; T, primary tumor category.
Smoking status was available for 51 patients.
Alcohol‐consumption data were available for 54 patients. Any reported consumption included former, occasional, or current alcohol use.
Documented comorbidity was defined as the presence of at least one comorbid condition; details regarding type and severity were not consistently available.
Initial treatment modality refers to the first recorded anticancer treatment and may not reflect the complete treatment sequence.
3.3. Treatment Completion and Tumor Response
Of 61 patients who initiated active treatment, 60 completed the prescribed course; one died from cardiac complications after the first induction chemotherapy cycle before formal response assessment. In the primary cohort, 55 of 56 patients completed treatment and were evaluable for response. Among evaluable patients, complete response, partial response, stable disease, and progressive disease occurred in 29.1%, 25.5%, 20.0%, and 25.5%, respectively, with no significant difference by CPS status (p = 0.636). Using all 56 patients as the denominator, ORR was 53.6% and DCR was 73.2%. Neither ORR (28.6% vs. 57.1%; p = 0.231) nor DCR (42.9% vs. 77.6%; p = 0.074) differed significantly between CPS groups (Supplemental Table 2).
3.4. Follow‐Up and Primary Survival Analyses
Median follow‐up was 89 months (95% CI, 51.5–126.5 months) in both the primary and entire cohorts. In the primary cohort, 47 deaths occurred. Median OS was 9 months (95% CI, 1.30–16.70) for CPS < 1 and 18 months (95% CI, 12.29–23.72) for CPS ≥ 1 (log‐rank p = 0.001; Figure 2A). CPS ≥ 1 was associated with lower mortality in univariable analysis (HR, 0.250; 95% CI, 0.104–0.598; p = 0.002) and after adjustment for age, ECOG performance status, T stage, and nodal status (adjusted HR, 0.243; 95% CI, 0.100–0.591; p = 0.002) (Table 2). T4a–T4b disease was independently associated with worse OS (adjusted HR, 2.537; 95% CI, 1.381–4.662; p = 0.003).
FIGURE 2.
Kaplan–Meier curves for overall survival and progression‐free survival according to PD‐L1 combined positive score status in the primary analytic cohort. (A) Median overall survival was 9 months in patients with CPS < 1 and 18 months in those with CPS ≥ 1 (log‐rank p = 0.001). (B) Median progression‐free survival was 5 months in patients with CPS < 1 and 12 months in those with CPS ≥ 1 (log‐rank p = 0.004). Vertical tick marks indicate censored observations. CPS, combined positive score; M0, absence of distant metastases; OS, overall survival; PD‐L1, programmed death‐ligand 1; PFS, progression‐free survival.
TABLE 2.
Kaplan–Meier and Cox regression analyses of overall survival and progression‐free survival according to PD‐L1 combined positive score status in the primary analytic cohort.
| Outcome | Events, n/N | CPS < 1, median months (95% CI) | CPS ≥ 1, median months (95% CI) | Log‐rank p value | Univariable HR (95% CI) a | p value | Adjusted HR (95% CI) b | p value |
|---|---|---|---|---|---|---|---|---|
| Overall survival | 47/56 | 9 (1.30–16.70) | 18 (12.29–23.72) | 0.001 | 0.250 (0.104–0.598) | 0.002 | 0.243 (0.100–0.591) | 0.002 |
| Progression‐free survival | 49/56 | 5 (2.66–7.34) | 12 (7.89–16.11) | 0.004 | 0.314 (0.133–0.742) | 0.008 | 0.356 (0.149–0.847) | 0.020 |
Note: Survival distributions were estimated using the Kaplan–Meier method and compared using the two‐sided log‐rank test. The primary analytic cohort comprised 56 patients without distant metastases who received active anticancer treatment. Overall survival events were deaths from any cause. Progression‐free survival events were documented disease progression or death from any cause.
Abbreviations: CI, confidence interval; CPS, combined positive score; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; PD‐L1, programmed death‐ligand 1.
Hazard ratios compare CPS ≥ 1 with CPS < 1; therefore, HR < 1 indicates lower hazard in the CPS ≥ 1 group.
Multivariable Cox models were adjusted for age, ECOG performance status, grouped T stage, and nodal status.
There were 49 PFS events. Median PFS was 5 months (95% CI, 2.66–7.34) for CPS < 1 and 12 months (95% CI, 7.89–16.11) for CPS ≥ 1 (log‐rank p = 0.004; Figure 2B). CPS ≥ 1 was associated with a lower hazard of progression or death in univariable analysis (HR, 0.314; 95% CI, 0.133–0.742; p = 0.008) and after adjustment (adjusted HR, 0.356; 95% CI, 0.149–0.847; p = 0.020) (Table 2). Advanced T stage was associated with worse PFS (adjusted HR, 2.165; 95% CI, 1.191–3.937; p = 0.011), whereas ECOG performance status ≥ 2 showed a borderline association (p = 0.055).
3.5. Additional and Sensitivity Analyses
In the primary cohort, the association of CPS ≥ 1 with lower hazards of OS and PFS remained significant after separate adjustment for documented comorbidity (OS: adjusted HR, 0.171; 95% CI, 0.064–0.453; p < 0.001; PFS: adjusted HR, 0.255; 95% CI, 0.100–0.653; p = 0.004) and treatment intent (OS: adjusted HR, 0.231; 95% CI, 0.093–0.569; p = 0.001; PFS: adjusted HR, 0.338; 95% CI, 0.141–0.808; p = 0.015). Recorded comorbidity was inversely associated with both outcomes, although this finding was limited by binary documentation and imbalance between CPS groups. This unexpected finding should be interpreted cautiously because comorbidity was recorded only as present or absent, its type and severity were unavailable, and all seven patients with CPS < 1 in the primary cohort had at least one documented comorbidity. Palliative intent was associated with worse OS, but estimates were imprecise because only four patients received palliative treatment (Supplemental Table 3).
Across the active‐treatment, M0, and entire cohorts, OS estimates remained directionally consistent (adjusted HRs, 0.292–0.455) and significant in most models. PFS associations were attenuated and nonsignificant or borderline in broader cohorts (adjusted HRs, 0.431–0.603; p = 0.051–0.212) (Table 3).
TABLE 3.
Primary and sensitivity analyses of the association between PD‐L1 CPS ≥ 1 and survival across analytic cohorts and adjustment models.
| Analysis | Cohort size, n (OS events/PFS events) | Adjustment variables | OS, adjusted HR (95% CI) | p | PFS, adjusted HR (95% CI) | p |
|---|---|---|---|---|---|---|
| Primary analytic cohort | 56 (47/49) | Age, ECOG PS, T stage, N status | 0.243 (0.100–0.591) | 0.002 | 0.356 (0.149–0.847) | 0.020 |
| Active‐treatment‐only cohort | 61 (52/54) | Age, ECOG PS, T stage, N status, M status | 0.292 (0.122–0.698) | 0.006 | 0.431 (0.185–1.005) | 0.051 |
| M0‐only cohort | 64 (55/57) | Age, ECOG PS, T stage, N status | 0.327 (0.144–0.741) | 0.007 | 0.462 (0.205–1.041) | 0.063 |
| M0‐only cohort, additionally adjusted for active treatment | 64 (55/57) | Age, ECOG PS, T stage, N status, active treatment | 0.343 (0.149–0.788) | 0.012 | 0.466 (0.206–1.056) | 0.067 |
| Entire study cohort | 71 (62/64) | Age, ECOG PS, T stage, N status, M status | 0.442 (0.200–0.977) | 0.044 | 0.600 (0.271–1.327) | 0.207 |
| Entire study cohort, additionally adjusted for active treatment | 71 (62/64) | Age, ECOG PS, T stage, N status, M status, active treatment | 0.455 (0.206–1.005) | 0.051 | 0.603 (0.273–1.335) | 0.212 |
| Entire study cohort, additionally adjusted for treatment intent | 71 (62/64) | Age, ECOG PS, T stage, N status, M status, treatment intent | 0.424 (0.192–0.933) | 0.033 | 0.545 (0.244–1.217) | 0.138 |
| Primary analytic cohort, additionally adjusted for comorbidity | 56 (47/49) | Age, ECOG PS, T stage, N status, comorbidity | 0.171 (0.064–0.453) | < 0.001 | 0.255 (0.100–0.653) | 0.004 |
| Primary analytic cohort, additionally adjusted for treatment intent | 56 (47/49) | Age, ECOG PS, T stage, N status, treatment intent | 0.231 (0.093–0.569) | 0.001 | 0.338 (0.141–0.808) | 0.015 |
Note: Hazard ratios compare CPS ≥ 1 with CPS < 1; HR < 1 indicates a lower hazard in the CPS ≥ 1 group. The primary analytic cohort included patients with M0 disease who received active anticancer treatment. The active‐treatment‐only cohort included patients with M0 or M1 disease but excluded those receiving supportive care only, whereas the M0‐only cohort included both actively treated patients and those receiving supportive care only. Active treatment was defined as receipt of any anticancer treatment versus supportive care only. Treatment intent was categorized as curative, palliative, or supportive care only in the entire cohort and as curative or palliative in the primary cohort. Treatment‐related estimates are associational and should not be interpreted as causal effects. Bold values indicate statistical significance (p < 0.05).
Abbreviations: CI, confidence interval; CPS, combined positive score; ECOG, Eastern Cooperative Oncology Group; HR, hazard ratio; M0, absence of distant metastases; M1, presence of distant metastases; N, regional lymph node category; OS, overall survival; PD‐L1, programmed death‐ligand 1; PFS, progression‐free survival; PS, performance status; T, primary tumor category.
3.6. Alternative Modeling of PD‐L1 CPS
In the primary cohort, CPS analyzed continuously was not associated with OS (adjusted HR per one‐point increase, 0.987; 95% CI, 0.971–1.004; p = 0.123) or PFS (adjusted HR, 0.993; 95% CI, 0.978–1.008; p = 0.376). Similarly, CPS ≥ 20 was not associated with OS (adjusted HR, 0.581; 95% CI, 0.302–1.117; p = 0.103) or PFS (adjusted HR, 0.636; 95% CI, 0.343–1.178; p = 0.150).
With CPS categorized as < 1, 1–19, and ≥ 20, median OS was 9, 14, and 19 months, respectively (global multivariable p = 0.005), and median PFS was 5, 11, and 14 months (global p = 0.045). Compared with CPS < 1, both CPS‐positive groups had lower mortality, whereas for PFS the association was significant for CPS ≥ 20 and borderline for CPS 1–19. Direct comparisons showed no significant differences between CPS ≥ 20 and CPS 1–19 for OS or PFS. Thus, the three‐category pattern primarily reflected poorer outcomes in the seven patients with CPS < 1 and should be interpreted cautiously (Supplemental Table 4).
3.7. Model Diagnostics
No evidence of proportional‐hazards violations was detected for CPS status, age, ECOG performance status, T stage, or nodal status in the primary OS or PFS models. The quadratic age term was nonsignificant in both models; therefore, age was retained as a linear continuous covariate (Supplemental Table 5).
4. Discussion
The principal finding was that PD‐L1 CPS ≥ 1 was associated with longer OS and PFS in the primary analytic cohort of actively treated M0 patients with HPSCC who received no immunotherapy. Median OS was 18 versus 9 months (adjusted HR, 0.243; 95% CI, 0.100–0.591), and median PFS was 12 versus 5 months (adjusted HR, 0.356; 95% CI, 0.149–0.847) for CPS ≥ 1 versus CPS < 1. These cohort‐specific findings suggest a possible prognostic association but do not establish CPS as a validated biomarker.
The OS association remained directionally consistent across broader cohorts and additional adjustment models, whereas the PFS association was attenuated and generally lost statistical significance. Restriction to actively treated M0 patients reduced but did not eliminate confounding related to disease extent, general condition, comorbidity, and treatment selection. The inverse association with recorded comorbidity should not be interpreted as protective because comorbidity was recorded only as present or absent and all seven patients with CPS < 1 had at least one comorbidity. Estimates for palliative intent were also imprecise because only four patients received palliative treatment.
Continuous CPS and the CPS ≥ 20 threshold were not associated with either outcome. Three‐category analyses suggested that the binary finding was driven mainly by poorer outcomes in the small CPS < 1 subgroup rather than by a dose–response relationship. Thus, the prognostic signal appeared threshold‐dependent rather than progressively stronger with increasing CPS values, and its robustness was greater for OS than for PFS. Objective response and disease control were numerically more frequent with CPS ≥ 1, but neither comparison was significant, and best‐response distributions did not differ. CPS was therefore not established as a marker of treatment sensitivity; its more consistent association with OS may instead reflect broader tumor–host characteristics. PD‐L1 expression is dynamic and may change during chemotherapy or radiotherapy through treatment‐related alterations in inflammatory signaling and the tumor immune microenvironment. In the present study, CPS was assessed exclusively on pretreatment biopsies and was therefore not influenced by subsequent therapy. Because paired post‐treatment specimens were unavailable, longitudinal changes in CPS could not be assessed.
Previous studies indicate that PD‐L1 is not a uniform prognostic marker across HNSCC sites and treatment settings. Meta‐analyses found no consistent association with OS, DFS, or DSS, although favorable associations with PFS or disease‐free survival were reported in some subgroups; heterogeneity in tumor site, antibody, cutoff, scoring method, tissue source, and the cellular compartment assessed likely contributes to these discrepancies [19, 20, 21]. Studies separating tumor‐cell from immune‐cell expression suggest that this distinction is important: PD‐L1 positivity on immune cells was associated with favorable outcomes and higher TIL density in resected or HPV‐negative HNSCC, whereas tumor‐cell PD‐L1 was less consistently prognostic [21, 22, 23]. Site‐specific findings further complicate interpretation, as PD‐L1 was associated with poorer outcomes in oral cavity cancer, while PD‐1‐positive TILs were favorable in oropharyngeal and hypopharyngeal tumors [24]. Other HNSCC cohorts have likewise reported either no association between 22C3 CPS and survival or favorable associations using different scoring approaches, underscoring that results are not directly interchangeable across assays and populations [25, 26].
HPSCC‐specific studies similarly support an immune‐context‐dependent interpretation. PD‐L1 expression has been linked to high TIL infiltration, but PD‐L1 alone was not consistently associated with survival; combined PD‐L1 positivity with high CD8 expression and CD8‐positive TIL density showed stronger prognostic relevance and, in some studies, associations with treatment response [27, 28]. Conversely, PD‐L1 has also been associated with advanced clinicopathological features, poorer survival, and tumor‐intrinsic pro‐aggressive signaling, whereas other studies found higher PD‐L1 expression and CD4‐positive infiltration to be independently associated with favorable OS [29, 30]. Additional work in laryngeal, pharyngeal, and hypopharyngeal tumors has shown that CD3‐, CD8‐, CD57‐, and FoxP3‐related immune‐cell densities or ratios may carry prognostic information, while PD‐L1 itself has shown variable or mainly univariable associations [31, 32, 33, 34]. Together, these findings suggest that the prognostic meaning of PD‐L1 in HPSCC depends less on expression alone than on the broader immune microenvironment and the method used to quantify it.
The favorable association between PD‐L1 CPS and survival without immunotherapy may appear counterintuitive, but PD‐L1 can reflect either tumor‐intrinsic immune evasion or adaptive immune resistance arising from a preexisting antitumor response. In the latter setting, inflammatory signaling induces PD‐L1 at the tumor–immune interface, particularly in TIL‐rich tumors [35, 36]. Because CPS includes PD‐L1‐positive tumor cells, lymphocytes, and macrophages, CPS ≥ 1 may therefore partly indicate an inflamed immune microenvironment. In HNSCC, immune‐cell PD‐L1 positivity and its combination with high CD8+ or FOXP3+ TIL infiltration have been associated with higher TIL density and favorable outcomes [21, 22, 23]. HPSCC‐specific studies likewise suggest that CD8+ TIL density, TIL‐subset ratios, and CD4+ infiltration may provide prognostic information beyond PD‐L1 expression alone [27, 28, 29, 32, 33]. This interpretation remains speculative because immune‐cell density and phenotype were not assessed, and CPS cannot distinguish the cellular source or spatial distribution of staining. Higher CPS should therefore be viewed as a possible surrogate of immune activation, not evidence of a specific mechanism.
From a clinical perspective, CPS is an established treatment‐selection biomarker in recurrent or metastatic HNSCC, where pembrolizumab‐based therapy improved OS, particularly in the CPS ≥ 1 and CPS ≥ 20 populations [13, 37]. Because no patient in the present cohort received immunotherapy, however, our findings concern the potential prognostic rather than predictive relevance of CPS. The association between CPS ≥ 1 and longer OS may reflect a more immune‐active tumor microenvironment, consistent with studies linking PD‐L1, particularly immune‐cell‐associated expression, to pre‐existing antitumor activity [21, 22, 27, 36, 38, 39]. Nevertheless, these cohort‐specific findings are hypothesis‐generating and do not support CPS as a standalone treatment‐selection marker. PD‐L1 assessment is affected by tumor heterogeneity, assay and scoring variability, and sampling site [14, 15, 16], while the small CPS < 1 group, absence of significant continuous or CPS ≥ 20 associations, and nonsignificant response comparisons further limit clinical interpretation. CPS should therefore be considered, at most, alongside established clinical and pathological prognostic factors [21, 22, 27, 40]. For CPS to contribute reliably to prognostic assessment or treatment selection, pretreatment tissue sampling, fixation and processing, tumor adequacy, assay performance, and CPS interpretation should be standardized. Its clinical use should also be based on integration with established clinical and pathological factors and supported by prospective validation in larger independent cohorts.
The study has several strengths. It focused specifically on HPSCC and excluded immunotherapy exposure. PD‐L1 was assessed on pretreatment primary‐tumor biopsies processed uniformly and evaluated using the validated 22C3 pharmDx assay by two pathologists blinded to outcomes, with discrepancies resolved by consensus. Robustness was examined across alternative cohorts, adjustment models, and CPS specifications, while model diagnostics supported the proportional hazards assumption and linear treatment of age. These features improve internal consistency but do not overcome limitations related to retrospective design and sample size.
This study also has several limitations. Its retrospective, single‐center design may introduce selection bias and limit generalizability. No formal sample‐size calculation was performed, and the primary cohort included 56 patients, only seven with CPS < 1, resulting in imprecise effect estimates and limited power for treatment‐response, subgroup, and alternative‐threshold analyses. This small subgroup therefore limits the precision and generalizability of the observed prognostic association.
Treatment was heterogeneous and reflected disease extent, performance status, comorbidities, anesthesiological risk, patient preference, and multidisciplinary judgment. Primary‐cohort restriction and sensitivity analyses reduced some heterogeneity, but residual confounding and confounding by indication remain possible. Smoking and alcohol data were incomplete, comorbidity was recorded only as present or absent, p16/HPV status was unavailable, and reliable cause‐of‐death data were lacking.
CPS was assessed on a single pretreatment biopsy, which may not capture spatial or temporal heterogeneity. Prolonged archival storage of FFPE tissue may affect PD‐L1 immunoreactivity, although published findings are inconsistent. Reduced 22C3 staining has been reported in older specimens, potentially leading to underestimation of PD‐L1 expression. Uniform processing, assay controls, and predefined adequacy criteria reduced this risk in our study, although an effect of storage duration cannot be excluded. No formal interobserver agreement coefficient was available, 9.9% of cases required consensus review, and response assessment was retrospective. Exploratory CPS and sensitivity analyses lacked multiplicity adjustment and do not establish an optimal threshold. Institutional biopsy requirements may have introduced further selection bias, and the absence of immunotherapy precludes conclusions about predictive value. External validation is required.
5. Conclusions
In this retrospective single‐center cohort of patients with hypopharyngeal squamous cell carcinoma treated without surgery or immunotherapy, CPS ≥ 1 was associated with longer OS and PFS in the primary analytic cohort. The OS association remained directionally consistent across sensitivity analyses, whereas the PFS association was less robust in broader cohorts. Exploratory analyses showed no clear linear association between increasing CPS and survival and no distinct association at the CPS ≥ 20 threshold, suggesting that the findings may largely reflect poorer outcomes in the CPS < 1 group. These cohort‐specific results do not establish CPS as a validated prognostic biomarker or support its use for treatment selection. Validation in larger, prospectively characterized cohorts is required.
Funding
The authors have nothing to report.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Supplemental Table 1. Baseline demographic, clinical, pathological, and treatment characteristics of the entire study cohort according to PD‐L1 combined positive score status.
Supplemental Table 2. Treatment completion and tumor response in the primary analytic cohort according to PD‐L1 combined positive score status.
Supplemental Table 3. Additional multivariable Cox regression models separately including documented comorbidity or treatment intent in the primary analytic cohort.
Supplemental Table 4. Alternative modeling of PD‐L1 combined positive score and its association with overall and progression‐free survival in the primary analytic cohort.
Supplemental Table 5. Assessment of the proportional hazards assumption and the functional form of age in the primary multivariable Cox regression models.
Acknowledgments
No artificial intelligence (AI) tools were used in the preparation of this manuscript.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- 1. Garneau J. C., Bakst R. L., and Miles B. A., “Hypopharyngeal Cancer: A State of the Art Review,” Oral Oncology 86 (2018): 244–250. [DOI] [PubMed] [Google Scholar]
- 2. Bradley P. J., “Epidemiology of Hypopharyngeal Cancer,” Advances in Oto‐Rhino‐Laryngology 83 (2019): 1–14. [DOI] [PubMed] [Google Scholar]
- 3. Newman J. R., Connolly T. M., Illing E. A., Kilgore M. L., Locher J. L., and Carroll W. R., “Survival Trends in Hypopharyngeal Cancer: A Population‐Based Review,” Laryngoscope 125, no. 3 (2015): 624–629. [DOI] [PubMed] [Google Scholar]
- 4. Kuo P., Chen M. M., Decker R. H., Yarbrough W. G., and Judson B. L., “Hypopharyngeal Cancer Incidence, Treatment, and Survival: Temporal Trends in the United States,” Laryngoscope 124, no. 9 (2014): 2064–2069. [DOI] [PubMed] [Google Scholar]
- 5. de Bree R., “The Current Indications for Non‐Surgical Treatment of Hypopharyngeal Cancer,” Advances in Oto‐Rhino‐Laryngology 83 (2019): 76–89. [DOI] [PubMed] [Google Scholar]
- 6. Eckel H. E. and Bradley P. J., “Treatment Options for Hypopharyngeal Cancer,” Advances in Oto‐Rhino‐Laryngology 83 (2019): 47–53. [DOI] [PubMed] [Google Scholar]
- 7. Chung E. J., Jeong W. J., Jung Y. H., et al., “Long‐Term Oncological and Functional Outcomes of Induction Chemotherapy Followed by (Chemo)radiotherapy vs Definitive Chemoradiotherapy vs Surgery‐Based Therapy in Locally Advanced Stage III/IV Hypopharyngeal Cancer: Multicenter Review of 266 Cases,” Oral Oncology 89 (2019): 84–94. [DOI] [PubMed] [Google Scholar]
- 8. Cui J., Wang L., Piao J., et al., “Initial Surgical Versus Non‐Surgical Treatments for Advanced Hypopharyngeal Cancer: A Meta‐Analysis With Trial Sequential Analysis,” International Journal of Surgery 82 (2020): 249–259. [DOI] [PubMed] [Google Scholar]
- 9. Habib A., “Management of Advanced Hypopharyngeal Carcinoma: Systematic Review of Survival Following Surgical and Non‐Surgical Treatments,” Journal of Laryngology and Otology 132, no. 5 (2018): 385–400. [DOI] [PubMed] [Google Scholar]
- 10. Lefebvre J. L., Andry G., Chevalier D., et al., “Laryngeal Preservation With Induction Chemotherapy for Hypopharyngeal Squamous Cell Carcinoma: 10‐Year Results of EORTC Trial 24891,” Annals of Oncology 23, no. 10 (2012): 2708–2714. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Juloori A., Koyfman S. A., Geiger J. L., et al., “Definitive Chemoradiation in Locally Advanced Squamous Cell Carcinoma of the Hypopharynx: Long‐Term Outcomes and Toxicity,” Anticancer Research 38, no. 6 (2018): 3543–3549. [DOI] [PubMed] [Google Scholar]
- 12. Ferris R. L., GJr B., Fayette J., et al., “Nivolumab vs Investigator's Choice in Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck: 2‐Year Long‐Term Survival Update of CheckMate 141 With Analyses by Tumor PD‐L1 Expression,” Oral Oncology 81 (2018): 45–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Burtness B., Harrington K. J., Greil R., et al., “Pembrolizumab Alone or With Chemotherapy Versus Cetuximab With Chemotherapy for Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck (KEYNOTE‐048): A Randomised, Open‐Label, Phase 3 Study,” Lancet 394, no. 10212 (2019): 1915–1928. [DOI] [PubMed] [Google Scholar]
- 14. Paolino G., Pantanowitz L., Barresi V., et al., “PD‐L1 Evaluation in Head and Neck Squamous Cell Carcinoma: Insights Regarding Specimens, Heterogeneity and Therapy,” Pathology, Research and Practice 226 (2021): 153605. [DOI] [PubMed] [Google Scholar]
- 15. Crosta S., Boldorini R., Bono F., et al., “PD‐L1 Testing and Squamous Cell Carcinoma of the Head and Neck: A Multicenter Study on the Diagnostic Reproducibility of Different Protocols,” Cancers (Basel) 13, no. 2 (2021): 292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. de Ruiter E. J., Mulder F. J., Koomen B. M., et al., “Comparison of Three PD‐L1 Immunohistochemical Assays in Head and Neck Squamous Cell Carcinoma (HNSCC),” Modern Pathology 34, no. 6 (2021): 1125–1132. [DOI] [PubMed] [Google Scholar]
- 17. Paintal A. S. and Brockstein B. E., “PD‐L1 CPS Scoring Accuracy in Small Biopsies and Aspirate Cell Blocks From Patients With Head and Neck Squamous Cell Carcinoma,” Head and Neck Pathology 14, no. 3 (2020): 657–665. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Emancipator K., Huang L., Aurora‐Garg D., et al., “Comparing Programmed Death Ligand 1 Scores for Predicting Pembrolizumab Efficacy in Head and Neck Cancer,” Modern Pathology 34, no. 3 (2021): 532–541. [DOI] [PubMed] [Google Scholar]
- 19. Tang H., Zhou X., Ye Y., Zhou Y., Wu C., and Xu Y., “The Different Role of PD‐L1 in Head and Neck Squamous Cell Carcinomas: A Meta‐Analysis,” Pathology, Research and Practice 216, no. 1 (2020): 152768. [DOI] [PubMed] [Google Scholar]
- 20. Yang W. F., Wong M. C. M., Thomson P. J., Li K. Y., and Su Y. X., “The Prognostic Role of PD‐L1 Expression for Survival in Head and Neck Squamous Cell Carcinoma: A Systematic Review and Meta‐Analysis,” Oral Oncology 86 (2018): 81–90. [DOI] [PubMed] [Google Scholar]
- 21. Lenouvel D., Gonzalez‐Moles M. A., Talbaoui A., et al., “An Update of Knowledge on PD‐L1 in Head and Neck Cancers: Physiologic, Prognostic and Therapeutic Perspectives,” Oral Diseases 26, no. 3 (2020): 511–526. [DOI] [PubMed] [Google Scholar]
- 22. Birtalan E., Danos K., Gurbi B., et al., “Expression of PD‐L1 on Immune Cells Shows Better Prognosis in Laryngeal, Oropharygeal, and Hypopharyngeal Cancer,” Applied Immunohistochemistry & Molecular Morphology 26, no. 7 (2018): e79–e85. [DOI] [PubMed] [Google Scholar]
- 23. Sanchez‐Canteli M., Granda‐Diaz R., Del Rio‐Ibisate N., et al., “PD‐L1 Expression Correlates With Tumor‐Infiltrating Lymphocytes and Better Prognosis in Patients With HPV‐Negative Head and Neck Squamous Cell Carcinomas,” Cancer Immunology, Immunotherapy 69, no. 10 (2020): 2089–2100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Schneider S., Kadletz L., Wiebringhaus R., et al., “PD‐1 and PD‐L1 Expression in HNSCC Primary Cancer and Related Lymph Node Metastasis – Impact on Clinical Outcome,” Histopathology 73, no. 4 (2018): 573–584. [DOI] [PubMed] [Google Scholar]
- 25. Wusiman D., Guo L., Huang Z., et al., “The Clinicopathological Significance of PD‐L1 Expression Assessed by the Combined Positive Score (CPS) in Head and Neck Squamous Cell Carcinoma,” Pathology, Research and Practice 236 (2022): 153934. [DOI] [PubMed] [Google Scholar]
- 26. Chen S. W., Li S. H., Shi D. B., et al., “Expression of PD‐1/PD‐L1 in Head and Neck Squamous Cell Carcinoma and Its Clinical Significance,” International Journal of Biological Markers 34, no. 4 (2019): 398–405. [DOI] [PubMed] [Google Scholar]
- 27. Hu C., Tian S., Lin L., Zhang J., and Ding H., “Prognostic and Clinicopathological Significance of PD‐L1 and Tumor Infiltrating Lymphocytes in Hypopharyngeal Squamous Cell Carcinoma,” Oral Oncology 102 (2020): 104560. [DOI] [PubMed] [Google Scholar]
- 28. Ono T., Azuma K., Kawahara A., et al., “Association Between PD‐L1 Expression Combined With Tumorinfiltrating Lymphocytes and the Prognosis of Patients With Advanced Hypopharyngeal Squamous Cell Carcinoma,” Oncotarget 8, no. 54 (2017): 92699–92714. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Shen L. F., Zhou S. H., and Guo Y., “Role of GLUT‐1 in the Upregulation of PD‐L1 Expression After Radiotherapy and Association of PD‐L1 With Favourable Overall Survival in Hypopharyngeal Cancer,” Oncotargets and Therapy 13 (2020): 11221–11235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Cui P., Jing P., Liu X., and Xu W., “Prognostic Significance of PD‐L1 Expression and Its Tumor‐Intrinsic Functions in Hypopharyngeal Squamous Cell Carcinoma,” Cancer Management and Research 12 (2020): 5893–5902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Karpathiou G., Casteillo F., Giroult J. B., et al., “Prognostic Impact of Immune Microenvironment in Laryngeal and Pharyngeal Squamous Cell Carcinoma: Immune Cell Subtypes, Immuno‐Suppressive Pathways and Clinicopathologic Characteristics,” Oncotarget 8, no. 12 (2017): 19310–19322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Wang J., Tian S., Sun J., Zhang J., Lin L., and Hu C., “The Presence of Tumour‐Infiltrating Lymphocytes (TILs) and the Ratios Between Different Subsets Serve as Prognostic Factors in Advanced Hypopharyngeal Squamous Cell Carcinoma,” BMC Cancer 20, no. 1 (2020): 731. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Landin D., Ahrlund‐Richter A., Mirzaie L., et al., “Immune Related Proteins and Tumor Infiltrating CD8+ Lymphocytes in Hypopharyngeal Cancer in Relation to Human Papillomavirus (HPV) and Clinical Outcome,” Head & Neck 42, no. 11 (2020): 3206–3217. [DOI] [PubMed] [Google Scholar]
- 34. Ou D., Adam J., Garberis I., et al., “Clinical Relevance of Tumor Infiltrating Lymphocytes, PD‐L1 Expression and Correlation With HPV/p16 in Head and Neck Cancer Treated With Bio‐ Or Chemo‐Radiotherapy,” Oncoimmunology 6, no. 9 (2017): e1341030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Scognamiglio T. and Yao‐Tseng C., “Beyond the Percentages of PD‐L1‐Positive Tumor Cells: Induced Versus Constitutive PD‐L1 Expression in Primary and Metastatic Head and Neck Squamous Cell Carcinoma,” Head and Neck Pathology 12, no. 2 (2018): 221–229. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Teng M. W. L., Ngiow S. F., Ribas A., and Smyth M. J., “Classifying Cancers Based on T‐Cell Infiltration and PD‐L1,” Cancer Research 75, no. 11 (2015): 2139–2145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Harrington K. J., Burtness B., Greil R., et al., “Pembrolizumab With or Without Chemotherapy in Recurrent or Metastatic Head and Neck Squamous Cell Carcinoma: Updated Results of the Phase III KEYNOTE‐048 Study,” Journal of Clinical Oncology 41, no. 4 (2023): 790–802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Bellmunt J., Mullane S. A., Werner L., et al., “Association of PD‐L1 Expression on Tumor‐Infiltrating Mononuclear Cells and Overall Survival in Patients With Urothelial Carcinoma,” Annals of Oncology 26, no. 4 (2015): 812–817. [DOI] [PubMed] [Google Scholar]
- 39. Mandai M., Hamanishi J., Abiko K., Matsumura N., Baba T., and Konishi I., “Dual Faces of IFN‐Gamma in Cancer Progression: A Role of PD‐L1 Induction in the Determination of Pro‐ and Anti‐Tumor Immunity,” Clinical Cancer Research 22, no. 10 (2016): 2329–2334. [DOI] [PubMed] [Google Scholar]
- 40. Papaioannou N. E., Beniata O. V., Vitsos P., Tsitsilonis O., and Samara P., “Harnessing the Immune System to Improve Cancer Therapy,” Annals of Translational Medicine 4, no. 14 (2016): 261. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental Table 1. Baseline demographic, clinical, pathological, and treatment characteristics of the entire study cohort according to PD‐L1 combined positive score status.
Supplemental Table 2. Treatment completion and tumor response in the primary analytic cohort according to PD‐L1 combined positive score status.
Supplemental Table 3. Additional multivariable Cox regression models separately including documented comorbidity or treatment intent in the primary analytic cohort.
Supplemental Table 4. Alternative modeling of PD‐L1 combined positive score and its association with overall and progression‐free survival in the primary analytic cohort.
Supplemental Table 5. Assessment of the proportional hazards assumption and the functional form of age in the primary multivariable Cox regression models.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.