Background
The KEYNOTE-048 study established the checkpoint inhibitor (CPI) pembrolizumab, with or without platinum chemotherapy, as standard frontline therapy for patients with recurrent/metastatic (R/M) head and neck squamous cell carcinoma (HNSCC). This study, along with the subsequent June 2019 FDA approval and revised NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines) for Head and Neck Cancers, established the importance of PD-L1 combined positive score (CPS) testing to guide frontline treatment selection for R/M HNSCC.1–3 Although pembrolizumab with chemotherapy was associated with improved overall survival (OS) compared with the EXTREME regimen, regardless of PD-L1 status, an OS benefit with pembrolizumab monotherapy was observed only in patients with PD-L1 CPS-positive tumors.4 For patients with PD-L1 CPS-negative disease, the response rate with pembrolizumab monotherapy was <5%, and survival outcomes were not superior to the EXTREME regimen.1,5 Thus, pembrolizumab monotherapy is recommended only for patients with PD-L1 CPS-positive disease.6
Despite the recommendation in the NCCN Guidelines, PD-L1 CPS testing is not universally performed,7 and the rate of treatment selection informed by PD-L1 testing remains unknown.8 Although pembrolizumab monotherapy lacks benefit in PD-L1–negative disease, clinicians may still prefer CPI monotherapy for some patients with R/M HNSCC, regardless of PD-L1 status, particularly those who are frail or elderly (aged ≥75 years), due to the perceived tolerability of single-agent CPI. Patterns of PD-L1 testing and factors associated with treatment selection have not been studied in a contemporary national dataset. Therefore, we aimed to investigate treatment patterns for R/M HNSCC in the United States in recent years, specifically examining overall uptake of CPI-based therapy, PD-L1 CPS testing rates, and “off-label” use of CPI monotherapy in patients with negative or unknown PD-L1 CPS.
Patients and Methods
Study Sample and Subsets
We included adult patients from the Flatiron Health electronic medical record (EHR)–derived deidentified nationwide database who initiated treatment for advanced HNSCC between January 13, 2011, and September 28, 2023. The Flatiron Health database is a longitudinal resource comprising deidentified patient-level structured and unstructured data, curated via technology-enabled abstraction.9,10 During the study period, the dataset included data from approximately 280 US cancer clinics (∼800 sites of care). All data are deidentified and subject to strict confidentiality safeguards to prevent reidentification and protect patient privacy. PD-L1 expression data in this dataset are captured from electronic health records from multiple sources, including laboratory and pathology reports, as well as clinician notes and other documents.
The overall cohort included patients who received systemic therapy in the frontline R/M setting, classified as chemotherapy ± cetuximab, CPI monotherapy, or CPI with chemotherapy. To avoid the misclassification of systemic therapy in the nonmetastatic setting (definitive or adjuvant chemoradiation) as systemic therapy in the metastatic setting, we defined frontline therapy as the first administration of a systemic treatment regimen at least 60 days after radiation to the primary site following a diagnosis of advanced HNSCC. To study CPI use in a contemporary cohort, we constructed a “2019+ cohort” comprising patients who started CPI after the 2019 FDA approval of frontline pembrolizumab. An institutional waiver of consent was granted by the University of Pennsylvania Institutional Review Board.
Study Measures
Patient characteristics recorded included age at frontline therapy initiation, sex (male, female), race (White, Black/African American, Asian, other/unknown), smoking history (current/former smoker, no smoking history), PD-L1 (CPS <1, 1–19, ≥20, unknown), ECOG performance status (PS; range, 0–4), primary cancer site (HPV-positive oropharynx, HPV-negative oropharynx, larynx, oral cavity, hypopharynx, unknown primary), year of frontline therapy, socioeconomic status (SES; 5-level indicator of neighborhood socioeconomic conditions, with 1 being the lowest and 5 being the highest11,12), practice setting (community, academic), and frontline treatment type (chemotherapy ± cetuximab, CPI with chemotherapy, CPI monotherapy). OS was defined as the time from frontline systemic therapy initiation to death or censored at last follow-up.
For patients in the 2019+ cohort, we defined “off-label” CPI monotherapy use as receipt of CPI monotherapy in the frontline R/M setting in patients who (1) had unknown or negative (<1) PD-L1 CPS prior to CPI treatment initiation, and (2) did not have platinum-based treatment within the prior 6 months before CPI initiation (ie, as part of definitive or adjuvant chemoradiotherapy with rapid progression indicating platinum refractoriness). The remainder of patients (ie, those receiving frontline chemoimmunotherapy or chemotherapy ± cetuximab, regardless of PD-L1 level, or those receiving CPI monotherapy with PD-L1 CPS ≥1 or within 6 months of prior platinum, regardless of PD-L1 level) were defined as “not off-label.”
Statistical Analysis
Baseline clinical and demographic characteristics of patients in the total cohort and the 2019+ cohort were analyzed using descriptive statistics. Continuous variables (age, SES) were summarized with median and IQR, whereas categorical variables (sex, race, smoking status, PD-L1 level, ECOG PS, HPV status, primary site, practice setting, type of frontline treatment) were summarized with frequency and percentage.
Distributions of frontline therapy by therapy type (CPI monotherapy, CPI with chemotherapy, chemotherapy ± cetuximab) over time, as well as by ECOG PS (0–1 vs ≥2) and age (≥75 vs <75 y), were examined graphically using frequencies and fractions of the total frequency for the 2019+ cohort. Treatment characteristics were compared between groups for categorical variables using Pearson’s chi-square test.
Within the 2019+ cohort, we calculated the proportion of patients receiving CPI monotherapy “off-label” and identified factors associated with “off-label” use using multivariable logistic regression. The following variables were included in the regression models: ECOG PS (0–1 vs ≥2), age (≥75 vs <75 years), sex (binary), race (White vs non-White), year of frontline treatment (continuous), smoking history (binary), practice setting (community vs academic), primary site (HPV-positive oropharynx, HPV-negative oropharynx, other), and SES (continuous). All variables with P<.10 in univariate logistic regression analysis were included in the full model. Backward elimination with a probability threshold for removal of 0.1 was then used to derive the final parsimonious model. Odds ratios (ORs) and 95% confidence intervals were used to represent the association between factors and the odds of “off-label” CPI monotherapy use. To address missing values in smoking history, ECOG PS, SES, and race, we used multiple imputation by chained equations (MICE). Fifty imputations were performed, using all covariates with and without missing data. Imputed results were then used to generate combined results of logistic regressions using Rubin’s rules.13 To detect any violation of the linearity assumption for the year of treatment variable, we fitted a restricted cubic spline model for year of treatment with 5 knots (2019, 2020, 2021, 2022, 2023) and examined a plot of logOR from another logistic regression model with year of treatment as categorical variable. The plot and the fitted model did not suggest a violation of the linearity assumption, and because only the linear term was significant, we reported models using the linear year term.
Finally, for the 2019+ cohort, OS from initiation of systemic therapy was estimated via Kaplan-Meier analysis, and log-rank test was used to compare survival between “off-label” and “not off-label” cohorts. A P<.05 was considered significant. Statistical analyses were performed using Stata, version 18.0 (StataCorp LLC).
Results
Total Cohort Characteristics
The final cohort comprised 7,657 patients with advanced HNSCC who received chemotherapy ± cetuximab (61%; n=4,675), CPI with chemotherapy (11%; n=864), or CPI monotherapy (28%; n=2,118) (Figure 1, Table 1). The median age was 65 years (IQR, 58–72), 77% were male, 67% were of White race, 78% had a history of smoking, 66% had ECOG PS 0–1, 69% were HPV-negative, and 76% were treated in a community setting. PD-L1 CPS was unknown prior to frontline therapy initiation in 73% of patients. Among patients with known PD-L1 status (n=2,052), distribution of CPS <1, 1–19, and ≥20 was 21% (n=437), 41% (n=840), and 38% (n=775), respectively. The most common primary cancer sites of the patients were HPV-positive oropharynx (25%), HPV-negative oropharynx (21%), larynx (21%), and oral cavity (25%).
Study inclusion diagram.
Abbreviations: CPI, checkpoint inhibitor; CPS, combined positive score; HNSCC, head and neck squamous cell carcinoma; R/M, recurrent/metastatic.
Citation: Journal of the National Comprehensive Cancer Network 23, 3; 10.6004/jnccn.2024.7085
Patient Demographics
| Total Cohort n (%) |
2019+ Cohorta | |||
|---|---|---|---|---|
| Total n (%) |
“Off-Label”b n (%) |
“Not Off-Label” n (%) | ||
| Total, N | 7,657 | 3,395 | 536 | 2,859 |
| Age, median (IQR), y | 65 (58–72) | 66 (59–73) | 68 (61–75) | 66 (59–73) |
| Sex | ||||
| Female | 1,728 (22.6) | 766 (22.6) | 129 (24.1) | 637 (22.3) |
| Male | 5,929 (77.4) | 2,629 (77.4) | 407 (75.9) | 2,222 (77.7) |
| Race | ||||
| White | 5,130 (67.0) | 2,192 (64.6) | 325 (60.6) | 1,867 (65.3) |
| Black/African American | 482 (6.3) | 186 (5.5) | 24 (4.5) | 162 (5.7) |
| Asian | 96 (1.3) | 38 (1.1) | 4 (0.7) | 34 (1.2) |
| Other | 1,234 (16.1) | 551 (16.2) | 118 (22.0) | 433 (15.1) |
| Missing | 715 (9.3) | 428 (12.6) | 65 (12.1) | 363 (12.7) |
| Smoking status | ||||
| No smoking history | 1,639 (21.4) | 814 (24.0) | 119 (22.2) | 695 (24.3) |
| Current/Former smoker | 5,983 (78.1) | 2,580 (76.0) | 417 (77.8) | 2,163 (75.7) |
| Missing | 35 (0.5) | 1 (0.0) | 0 (0.0) | 1 (0.0) |
| PD-L1 level | ||||
| CPS ≥20 | 775 (10.1) | 702 (20.7) | 0 (0.0) | 702 (24.6) |
| CPS 1–19 | 840 (11.0) | 767 (22.6) | 0 (0.0) | 767 (26.8) |
| CPS <1 | 437 (5.7) | 335 (9.9) | 87 (16.2) | 248 (8.7) |
| Unknownc | 5,605 (73.2) | 1,591 (46.9) | 449 (83.8) | 1,142 (39.9) |
| ECOG PS (ref: 0–1) | ||||
| 0–1 | 5,029 (65.7) | 2,434 (71.7) | 372 (69.4) | 2,062 (72.1) |
| ≥2 | 1,304 (17.0) | 571 (16.8) | 98 (18.3) | 473 (16.5) |
| Missing | 1,324 (17.3) | 390 (11.5) | 66 (12.3) | 324 (11.3) |
| Primary site | ||||
| HPV-positive oropharynx | 1,928 (25.2) | 1,051 (31.0) | 166 (31.0) | 885 (31.0) |
| HPV-negative oropharynx | 1,581 (20.6) | 543 (16.0) | 99 (18.5) | 444 (15.5) |
| Larynx | 1,567 (20.5) | 672 (19.8) | 127 (23.7) | 545 (19.1) |
| Oral cavity | 1,899 (24.8) | 869 (25.6) | 98 (18.3) | 771 (27.0) |
| Hypopharynx | 487 (6.4) | 204 (6.0) | 33 (6.2) | 171 (6.0) |
| Unknown primary | 195 (2.5) | 56 (1.6) | 13 (2.4) | 43 (1.5) |
| Year of treatment | ||||
| 2011–2018 | 4,262 (55.7) | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| 2019 | 765 (10.0) | 765 (22.5) | 186 (34.7) | 579 (20.3) |
| 2020 | 735 (9.6) | 735 (21.6) | 112 (20.9) | 623 (21.8) |
| 2021 | 736 (9.6) | 736 (21.7) | 109 (20.3) | 627 (21.9) |
| 2022 | 725 (9.5) | 725 (21.4) | 77 (14.4) | 648 (22.7) |
| 2023 | 434 (5.7) | 434 (12.8) | 52 (9.7) | 382 (13.4) |
| SES, median (IQR)d | 3 (2–4) | 3 (2–4) | 3 (2–4) | 3 (2–4) |
| Practice setting | ||||
| Academic | 1,819 (23.8) | 791 (23.3) | 97 (18.1) | 694 (24.3) |
| Community | 5,838 (76.2) | 2,604 (76.7) | 439 (81.9) | 2,165 (75.7) |
| Frontline treatment | ||||
| Chemotherapy ± cetuximab | 4,675 (61.1) | 1,102 (32.5) | 0 (0.0) | 1,102 (38.5) |
| CPI with chemotherapy | 864 (11.3) | 832 (24.5) | 0 (0.0) | 832 (29.1) |
| CPI monotherapy | 2,118 (27.7) | 1,461 (43.0) | 536 (100.0) | 925 (32.4) |
Abbreviations: CPI, checkpoint inhibitor; CPS, combined positive score; PS, performance status; SES, socioeconomic status.
Patients who started CPI treatment after the 2019 FDA approval of frontline pembrolizumab.
“Off-label” use of CPI monotherapy is defined as single-agent CPI (ie, no concurrent chemotherapy) administered in patients with negative or unknown PD-L1 CPS.
Includes number of patients with test attempted but no result available and patients who had first-line therapy initiation prior to PD-L1 testing.
Five-level indicator of neighborhood SES, with 1 being the lowest, and 5 being the highest.
Frontline Treatment Trend Over Time in Total Cohort
Within the total cohort (n=7,657), use of frontline CPI monotherapy and CPI with chemotherapy steadily increased over time compared with chemotherapy ± cetuximab (Figure 2). Starting in 2019, CPI-based treatment comprised more than half of all frontline treatment received; in 2023, CPI monotherapy (42%) and CPI with chemotherapy (36%) accounted for more than three-quarters of frontline therapy for R/M HNSCC.
Distribution of treatment type by year. Numbers above each bar indicate total number of treated patients. Numbers in each bar indicate the percentages for the corresponding treatments.
Abbreviation: CPI, checkpoint inhibitor.
Citation: Journal of the National Comprehensive Cancer Network 23, 3; 10.6004/jnccn.2024.7085
2019+ Subgroup Characteristics and Treatment Distributions
Of the total cohort, 3,395 patients were treated in 2019 and beyond (2019+ cohort) (Table 1). Demographics and disease characteristics were similar to those of the total cohort, with notable exceptions being that the percentage of patients missing PD-L1 levels was lower (47% vs 73%), and a majority of patients were treated with CPI-based therapy, as noted earlier.
Within the 2019+ subgroup, the most common frontline treatment was CPI monotherapy (43%), followed by chemotherapy ± cetuximab (33%) and CPI with chemotherapy (25%). CPI monotherapy was more common in patients aged ≥75 years (54%) compared with those aged <75 years (40%) (Pearson’s chi-square P<.001; Figure 3A). CPI monotherapy was also more common in those with ECOG PS ≥2 (52%) compared with ECOG PS 0–1 (41%) (Pearson’s chi-square P<.001; Figure 3B).
Distribution of treatment type by (A) age and (B) ECOG PS for the 2019+ cohort. Numbers above each bar indicate total number of treated patients. Numbers in each bar indicate the percentages for the corresponding treatments.
Abbreviations: CPI, checkpoint inhibitor; PS, performance status.
Citation: Journal of the National Comprehensive Cancer Network 23, 3; 10.6004/jnccn.2024.7085
Nearly half (47%) of the patients in the 2019+ cohort did not have a recorded PD-L1 test result prior to initiation of frontline R/M therapy. Among those with known PD-L1 status (n=1,804), the distribution of CPS was as follows: 19% (n=335) with CPS <1, 43% (n=767) with CPS 1–19, and 39% (n=702) with CPS ≥20 (Table 1). CPI monotherapy was the most common frontline treatment type for tumors with CPS ≥20 (56%), but it was also used in 49% of CPS 1–19, 33% of CPS <1, and 37% of CPS unknown (Figure 4). From 2019 to 2023, the use of CPI with chemotherapy generally increased over time, whereas the use of chemotherapy ± cetuximab and CPI monotherapy generally decreased (Supplementary Figure S1, available online in the supplementary materials).
Distribution of treatment type by PD-L1 status for the 2019+ cohort. Numbers above each bar indicate total number of treated patients. Numbers in each bar indicate the percentages for the corresponding treatments.
Abbreviations: CPI, checkpoint inhibitor; CPS, combined positive score.
Citation: Journal of the National Comprehensive Cancer Network 23, 3; 10.6004/jnccn.2024.7085
Factors Associated With “Off-Label” CPI Monotherapy Use
Among the 2019+ subgroup of patients with PD-L1 CPS negative or unknown status (n=1,926), 28% (n=536) received “off-label” CPI monotherapy. Among the remaining 1,390 patients with PD-L1 CPS negative or unknown status “not off-label,” 30.5% (n=424) received CPI + chemotherapy, 58.2% (n=809) received chemotherapy ± cetuximab, and 11.3% (n=157) received CPI monotherapy.
In univariate analysis, factors associated with “off-label” CPI monotherapy use included age, race, year of frontline treatment, and practice setting (at P=.1 significance level). These variables were included in the multivariate logistic regression model. After backward elimination, the final model identified 3 factors associated with “off-label” CPI monotherapy use: age ≥75 years (OR, 1.4; 95% CI, 1.1–1.7), earlier year of frontline treatment (OR, 1.3 per earlier year; 95% CI, 1.2–1.4), and community practice setting (OR, 1.4; 95% CI, 1.1–1.8) (Table 2).
Univariate and Multivariate Logistic Regression Analysis of “Off-Label” CPI Monotherapy Use in the 2019+ Cohorta (N=3,395)
| Univariate Analysis | Multivariate Analysis Final Modelb | |||
|---|---|---|---|---|
| OR (95% CI) | P Value | OR (95% CI) | P Value | |
| ECOG PS ≥2 (ref: 0–1) | 1.128 (0.884–1.439) | .332 | ||
| Age ≥75 y (ref: <75 y) | 1.296 (1.046–1.607) | .018 | 1.360 (1.094–1.689) | .006 |
| Male (ref: female) | 0.904 (0.728–1.123) | .364 | ||
| White race (ref: non-White) | 0.773 (0.625–0.956) | .017 | ||
| Earlier year of treatment (per year) | 1.284 (1.195–1.379) | <.001 | 1.291 (1.202–1.388) | <.001 |
| Current/Former smoker (ref: no smoking history) | 1.126 (0.903–1.404) | .294 | ||
| Community practice setting (ref: academic) | 1.451 (1.146–1.837) | .002 | 1.450 (1.143–1.839) | .002 |
| Primary site (ref: HPV-positive oropharynx) | ||||
| HPV-negative oropharynx | 1.189 (0.904–1.563) | .216 | ||
| Other | 0.944 (0.765–1.165) | .593 | ||
| SESc (per 1 unit increase of SES) | 0.988 (0.920–1.062) | .745 | ||
Bold indicates statistically significant P value.
Abbreviations: CPI, checkpoint inhibitor; OR, odds ratio; PS, performance status; SES, socioeconomic status.
Patients who started CPI treatment after the 2019 FDA approval of frontline pembrolizumab.
Logistic regression for odds of off-label CPI monotherapy use. All variables with P values <.10 in univariate logistic regression analysis were included in the full model. Backward elimination with a probability threshold for removal of 0.1 was then used to derive the final parsimonious model. Multiple imputation by chained equations (MICE) was used to account for missing values.
Five-level indicator of neighborhood SES, with 1 being the lowest and 5 being the highest.
Survival
In the 2019+ cohort, the median OS was 12.9 months, with no significant difference in survival between the “off-label” and “not off-label” cohorts (P=.376) (Supplementary Figure S2).
Discussion
In this US nationwide retrospective study of advanced HNSCC, use of CPI-based treatment has increased over the 5 years following the FDA approval of frontline pembrolizumab and has become the predominant form of frontline therapy in the R/M setting. Despite guidelines recommending that CPI monotherapy be reserved for patients with positive PD-L1 expression, nearly 50% of patients treated after 2019 were not tested for PD-L1 expression before starting therapy. Additionally, 28% of patients with negative or unknown PD-L1 CPS received CPI monotherapy in an “off-label” fashion.
There are several important considerations and potential explanations for our findings. First, the higher “off-label” use observed among elderly patients and those with ECOG PS ≥2 suggests that some “off-label” CPI monotherapy use may be occurring in patients who are ineligible for platinum or chemotherapy, either due to frailty or comorbidities. Indeed, prior studies have shown increasingly prevalent “off-label” use of immunotherapy (∼18%–30% across cancer types).14–16 Saleh et al17 reported even higher rates (∼49%) in patients aged ≥75 years. Similarly, Parikh et al18 found that patients ineligible for clinical trials due to frailty or organ dysfunction were preferentially prescribed immunotherapy monotherapy. The potential for durable benefit with immunotherapy may also bias providers in favor of CPI monotherapy, but it must be emphasized that only a minority of patients with R/M HNSCC, and certainly very few with negative PD-L1 expression, are likely to achieve long-term response or remission when compared with more immunotherapy-responsive cancers such as melanoma or non–small cell lung cancer.
These findings have important clinical implications. Although CPI monotherapy may be reasonable in chemotherapy-ineligible patients with negative/unknown PD-L1 CPS (although guidelines do not explicitly recommend CPI monotherapy in the platinum- or chemotherapy-ineligible setting), its use in patients with PD-L1–negative disease who are eligible for chemotherapy risks undertreatment with a less efficacious regimen. Notably, the lack of a statistically significant difference in OS between the “off-label” and “not off-label” groups in our cohort should not be interpreted as evidence that “off-label” CPI monotherapy is an acceptable or noninferior treatment strategy. Valid comparative efficacy conclusions cannot be drawn between these groups due to significant baseline differences in multiple recorded and unrecorded demographic, cancer-related, and treatment factors that cannot be easily adjusted for.
We observed that the community practice setting was associated with a 40% higher likelihood of “off-label” CPI monotherapy use compared with the academic practice setting, possibly reflecting delayed community uptake of biomarker testing and nuanced guideline recommendations in community settings, as has been observed in other cancer types.19–21 Encouragingly, “off-label” use of CPI monotherapy has declined steadily since 2019, perhaps indicating increased provider familiarity with the nuance of PD-L1–guided treatment selection and recognition of the inferior responses with CPI monotherapy in PD-L1–negative tumors.
Although prior studies have explored “off-label” use of CPI monotherapy in solid tumors in general, our study is the first to focus on its use in R/M HNSCC, using a contemporary cohort and a large, nationwide EHR-derived dataset. This dataset allowed us to analyze patterns of testing and treatment following KEYNOTE-048 and FDA approval and show that although “off-label” use remains common, it has shown a decline in recent years.
Inherent limitations of this retrospective analysis include incomplete data capture of variables such as race and PD-L1 CPS. It is possible that despite technology-enabled data abstraction from multiple sources,9 some patients receiving CPI monotherapy may have had a positive PD-L1 test that was not recorded in the database, leading to misclassification as “off-label.” In addition, we were unable to capture variables such as disease burden or comorbidities that could influence treatment selection, nor could we ascertain the exact reasons for “off-label” use.
Conclusions
This nationwide observational study demonstrates that most patients with R/M HNSCC are now receiving CPI in the frontline setting, but that PD-L1 CPS testing and PD-L1–guided treatment decisions remain suboptimal. “Off-label” use of PD-L1 monotherapy in PD-L1–negative or PD-L1–unknown HNSCC is more common in elderly and frail patients, as well as those treated in community practice settings, but has shown a declining trend in recent years.
References
- 1.↑
Ferris RL, Blumenschein G, Fayette J, et al. Nivolumab for recurrent squamous-cell carcinoma of the head and neck. N Engl J Med 2016;375:1856–1867.
- 2.↑
Chow LQM, Haddad R, Gupta S, et al. Antitumor activity of pembrolizumab in biomarker-unselected patients with recurrent and/or metastatic head and neck squamous cell carcinoma: results from the phase Ib KEYNOTE-012 expansion cohort. J Clin Oncol 2016;34:3838–3845.
- 3.↑
Pfister DG, Spencer S, Adkins D, et al. NCCN Clinical Practice Guidelines in Oncology: Head and Neck Cancer. Version 1.2025. To view the most recent version, visit https://www.nccn.org
- 4.↑
Burtness B, Harrington KJ, 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 2019;394:1915–1928.
- 6.↑
Harrington KJ, 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. J Clin Oncol 2023;41:790–802.
- 7.↑
Black CM, Hair G, Zheng D, et al. To test, or not to test, that is the question: a real-world analysis of PD-L1 expression testing patterns in recurrent/metastatic head and neck squamous cell carcinoma (R/M HNSCC). J Clin Oncol 2023;41(Suppl):Abstract 6033.
- 8.↑
Black CM, Hanna GJ, Wang L, et al. Real-world treatment patterns and outcomes among individuals receiving first-line pembrolizumab therapy for recurrent/metastatic head and neck squamous cell carcinoma. Front Oncol 2023;13:1240947.
- 9.↑
Ma X, Long L, Moon S, et al. Comparison of population characteristics in real-world clinical oncology databases in the US: Flatiron Health, SEER, and NPCR. medRxiv. Preprint posted online July 6, 2023. doi:10.1101/2020.03.16.20037143
- 10.↑
Birnbaum B, Nussbaum N, Seidl-Rathkopf K, et al. Model-assisted cohort selection with bias analysis for generating large-scale cohorts from the EHR for oncology research. arXiv. Preprint posted online January 13, 2020. doi:10.48550/arXiv.2001.09765
- 11.↑
Yu M, Tatalovich Z, Gibson JT, Cronin KA. Using a composite index of socioeconomic status to investigate health disparities while protecting the confidentiality of cancer registry data. Cancer Causes Control 2014;25:81–92.
- 12.↑
Yost K, Perkins C, Cohen R, et al. Socioeconomic status and breast cancer incidence in California for different race/ethnic groups. Cancer Causes Control 2001;12:703–711.
- 14.↑
Haslam A, Prasad V. Estimation of the percentage of US patients with cancer who are eligible for and respond to checkpoint inhibitor immunotherapy drugs. JAMA Netw Open 2019;2:e192535.
- 15.↑
Souza JAD, Duong YY. Off-label immunotherapy prescription: financial implications for payers and patients. J Clin Oncol 2017;35(Suppl):Abstract 6.
- 16.↑
Hsu JC, Lin JY, Hsu MY, Lin PC. Effectiveness and safety of immune checkpoint inhibitors: a retrospective study in Taiwan. PLoS One 2018;13:e0202725.
- 17.↑
Saleh K, Auperin A, Martin N, et al. Efficacy and safety of immune checkpoint inhibitors in elderly patients (≥70 years) with squamous cell carcinoma of the head and neck. Eur J Cancer 2021;157:190–197.
- 18.↑
Parikh RB, Min EJ, Wileyto EP, et al. Uptake and survival outcomes following immune checkpoint inhibitor therapy among trial-ineligible patients with advanced solid cancers. JAMA Oncol 2021;7:1843–1850.
- 19.↑
Sadik H, Pritchard D, Keeling DM, et al. Impact of clinical practice gaps on the implementation of personalized medicine in advanced non–small-cell lung cancer. JCO Precis Oncol 2022;6:e2200246.
- 20.↑
Barcellini A, Dal Mas F, Paoloni P, et al. Please mind the gap-about equity and access to care in oncology. ESMO Open 2021;6:100335.
- 21.↑
Mileham KF, Schenkel C, Bruinooge SS, et al. Defining comprehensive biomarker-related testing and treatment practices for advanced non-small-cell lung cancer: results of a survey of U.S. oncologists. Cancer Med 2022;11:530–538.
