Introduction

Immune checkpoint blockade (ICB) therapy represents a paradigm-shifting breakthrough in cancer care since the anti-cytotoxic T-lymphocyte associated protein 4 (CTLA-4) antibody ipilimumab was approved by the US Food and Drug Administration (FDA) in 2011 for metastatic melanoma. Numerous ICB therapies have since come to market, resulting in a dramatic change in the treatment landscape and improved progression-free and overall survival (PFS and OS, respectively) in several different cancers.1 The benefit of ICB, however, is not universal across cancer types or even among patients with the same malignancy. Significant research is underway to understand what controls response and resistance to ICB, but this is ultimately dependent on complex interactions between the immune system and tumor cells. Predicting which patients will have adverse events (AEs), and the manifestations of these, is also challenging.

Several biological factors can modulate the response to ICB. For example, sex hormones differences affect immune cell function.2 High expression of testosterone has been associated with decreased antibody production, increased proliferation of T regulatory (Treg) cells, decreased CD8+ T cell proliferation and secretion of cytokines, increased production of anti-inflammatory neutrophils, and reduced natural killer cell cytotoxicity.2 3 Estrogen levels are also associated with the immunomodulatory effects of adipose tissues. Obesity is considered a risk factor for cancer development but may increase the efficacy of ICBs.4 Studies have reported that populations of African ancestry have stronger transcriptional responses to immune challenge than those of European descent.5 Finally, there is increasing recognition that aging has a profound effect on the function of the immune system.6 Despite the mounting evidence that these and other factors modulate the function of the immune system, little is known about how they correlate with clinical outcomes after ICB therapy. Pivotal clinical trials represent an important mechanism for evidence generation for the efficacy and AEs of treatments. Adequate representation of patient populations and efforts to diversify and include groups historically under-represented in clinical research are needed to determine these correlations and to ensure the generalizability of trial results. Historically, clinical trials have over-represented white men,7 8 younger and healthier patients, individuals living near urban academic centers, and in high-income countries. Lack of adequate representation poses risks of reduced efficacy and increased AEs among under-represented populations.

In this review, we aim to characterize the representation of participants in the key phase 2 and 3 clinical trials of ICB therapy that inform current clinical practice among common solid and hematologic malignancies. We investigate whether the trials reported demographic information like age, sex, race and ethnicity, and possible biological confounders such as obesity/body mass index either in publications reporting trial results or on ClinicalTrials.gov. We also aimed to determine if outcomes and AEs differed among different patient subgroups. Importantly, while we will focus on key phase 2 and 3 trials, representation in early-stage clinical trials are also important as crucial scientific decisions that impact the dosing, design, and selection of treatment options rely on these studies.

Solid malignancies

ICB has become the standard of care in multiple solid malignancies and numerous trials are being conducted to gain new treatment indications. We will focus on selected landmark trials that led to the FDA approval of ICB. Relevant information from these trials is summarized in table 1 and online supplemental table 1.

Table 1

Select phase 2 and 3 landmark immunotherapy trials in solid malignancies

Skin cancers

Ipilimumab gained FDA approval in 2011 for refractory melanoma following the phase 3 MDX0100-020 trial, which showed superior median OS (mOS) versus glycoprotein 100 (gp100, 10 vs 6.4 months).9 Subgroup analysis showed that both patients <65 years of age and ≥65 had significant OS benefit with ipilimumab compared with gp100 alone, with HR of 0.65 and 0.61, respectively. Sex-based subgroup analysis revealed that males received more benefit from ipilimumab over gp100 (HR 0.54) than females (HR 0.81). KEYNOTE-006 demonstrated pembrolizumab’s superior PFS and OS over ipilimumab (7-year OS 37% vs 25%).10 In CheckMate 067, the combination of nivolumab plus ipilimumab showed an mOS of 72.1 months, more twice that of either agent alone.11 12 Subgroup analysis for OS showed no difference based on age and was not conducted by sex. Racial demographics were not reported in the publications.

In IMspire150, an improved PFS was observed with the addition of atezolizumab to vemurafenib and cobimetinib for BRAF V600 mutated melanoma (15.1 months) versus placebo with vemurafenib and cobimetinib (10.6 months), but the incidence of grade 3–4 AEs was higher.13 Subgroup analyses showed favorable outcomes in patients ≥65 years old (HR 0.63) compared with <65 (HR 0.76) and whites (HR 0.76) compared with non-white (HR 1.07). However, only 5% of the study participants were non-white. Similar PFS was observed in men (HR 0.75) and women (HR 0.81).

Pembrolizumab and cemiplimab are FDA-approved treatments for locally advanced/metastatic cutaneous squamous cell carcinoma. In phase 2 KEYNOTE-629 trial, pembrolizumab showed an overall response rate (ORR) of 34% and median PFS (mPFS) was 6.9 months.14 The study included mostly men (76%) and older adults, with a median age of 72 years. No subgroup analyses were performed.

KEYNOTE-017 was a phase 2 trial that evaluated pembrolizumab as frontline therapy for advanced Merkel cell carcinoma and showed an ORR of 56%, mPFS of 16.8 months and 24-month OS was 69%.15 It enrolled mostly men (68%) and older adults (80% ≥65 years old), however, no subgroup analysis was performed. The FDA approval for avelumab in this disease was based on the JAVELIN Merkel-200 trial.16 ORR was 31.8% in the study population, which were mostly men (74%) and older adults (75% ≥65 years old). Subgroup analysis performed by age, sex, and pooled region (North America, Europe and the rest of the world) to assess ORR showed no significant difference between the groups.

Gastrointestinal malignancies

The role of ICB gastroesophageal cancer centers on programmed death-ligand 1 (PD-L1) expression and tumor location. KEYNOTE-590 led to FDA approval of pembrolizumab plus chemotherapy for esophageal and gastroesophageal junction carcinomas.17 This trial showed improved OS (12.4 vs 9.8 months), with a 5-year update revealing a higher OS rate (11% vs 3%).18 The study enrolled mostly men (83.5%) and Asians (53%), with subgroup analyses favoring pembrolizumab across age groups, gender, and geography (Asian vs non-Asian) participants. In CheckMate 648, improved OS was observed with frontline nivolumab with chemotherapy or in combination with ipilimumab versus chemotherapy alone (24 months OS: 29% vs 32% vs 19%, respectively) in squamous carcinomas.19 20 The study also enrolled mostly Asians (71%) but race was not included in the subgroup analysis. CheckMate 649 evaluated nivolumab plus chemotherapy, nivolumab and ipilimumab, or chemotherapy in adenocarcinomas and showed improved OS for nivolumab plus chemotherapy in patients with high PD-L1 expression (14.4 vs 11.1 months).21 Subgroup analysis favored all age groups, men, and both white and Asian participants. CheckMate 577 showed improved disease-free survival (DFS) with neoadjuvant nivolumab versus placebo (median DFS 22.4 vs 11 months, HR 0.69). Subgroup analysis represented all demographic data reported including race and region. Among racial groups, median DFS was only statistically significant in white race (HR 0.71).22

Early ICB studies in metastatic colorectal cancer (mCRC) were discouraging, but promising responses emerged in patients with high microsatellite instability/deficient DNA mismatch repair (MSI-H/dMMR). KEYNOTE-177 compared pembrolizumab to chemotherapy with or without bevacizumab or cetuximab in MSI-H/dMMR mCRC and showed superior 3-year PFS and fewer AEs with pembrolizumab (42% vs 11%).23 Subgroup analysis favored pembrolizumab in all major subgroups including ≤70 years old, males and females, BRAF wild type and V600E mutated, and right-sided primary tumor, but was not significant (though still favored pembrolizumab) for >70 years old. In CheckMate 142, first-line nivolumab showed an ORR of 69%, with 51 patients achieving disease control for at least 12 weeks.24 Subgroup analysis revealed that <65 years old had a higher ORR (77%) than those ≥65 years old (61%), and males exhibited higher ORR (78%) compared with females (59%). No racial or ethnicity data analysis was reported.

Single-agent pembrolizumab initially showed effectiveness in treating hepatocellular carcinoma (HCC), leading to accelerated FDA approval.25 However, KEYNOTE-240, which compared pembrolizumab to placebo in second-line HCC, did not achieve statistical significance in its primary endpoints of PFS and OS.26 Advanced HCC now adopts standard combination strategies. The IMbrave-150 trial, for instance, demonstrated superior PFS (6.8 vs 4.3 months) and an mOS of 19.2 versus 13.4 months with atezolizumab-bevacizumab compared with sorafenib.27 28 The study enrolled mostly men (83%), with a median age of 65 years old, and of Asian race (57%), with 60% of participants enrolled outside of Asia. Subgroup analysis favored these predominant groups. However, atezolizumab-bevacizumab was superior than sorafenib regardless of geographic location. The HIMALAYA trial showed that tremelimumab plus durvalumab improved 4-year OS compared with sorafenib (25.2% vs 15.1%).29 This study enrolled mostly males, median age of 65 years old, and from non-Asia countries. Race was not reported. Subgroup analysis was reported for age, sex and geographic region and showed statistical significance in the elderly and males.

Genitourinary malignancies

Pembrolizumab gained first-line approval for cisplatin-ineligible, PD-L1 positive urothelial carcinomas based on data from KEYNOTE-045, which showed superior OS over chemotherapy in platinum-refractory disease (mOS 10.3 vs 7.4 months).30 CheckMate 901 showed improved OS with first-line nivolumab plus chemotherapy compared with chemotherapy alone (mOS 21.7 vs 18.9 months). Subgroup analysis did not show OS benefit for specific races or regions except for Europe (HR 0.73).31 In JAVELIN BLADDER-100, best supportive care plus avelumab demonstrated superior mOS (21.4 vs 14.3 months, HR 0.69) and PFS (3.7 vs 2.0 months, HR 0.62) for unresectable urothelial cancer.32 Most patients were male (67%), white (67%) and from Europe (60%), with a median age of 69 years old. OS was significant with avelumab in the predominating represented groups: male (HR 0.64), age >65 years old (HR 0.63), white (HR 0.67) and Europe region (HR 0.64).

Treatment options for metastatic renal cell carcinoma (RCC) rely heavily on ICB. KEYNOTE-426 supported the approval of axitinib plus pembrolizumab as first-line therapy for clear cell RCC, showing superior mPFS versus sunitinib (15 vs 11 months) and a higher ORR (59.3% vs 35.7%).33 This global study enrolled predominantly male (73%) with a median age of 61.5 years. Subgroup analysis showed favorable responses across all ages, sex, and regions. Analysis by race was not reported. CheckMate 214 compared nivolumab plus ipilimumab with sunitinib in advanced RCC and showed higher ORR (42% vs 27%) and longer mOS with nivolumab (47 vs 26.6 months).34 The study enrolled mostly males and had a median age of 62. Subgroup analyses favored nivolumab in both sexes, age <65 years old, and regions excluding Canada, Europe, and the USA. Adjuvant pembrolizumab yielded improvements in DFS when compared with placebo among patients with kidney cancer who were at high risk for recurrence (77.3% vs 68.1%; HR 0.68).35 Geographic region was reported in the subgroup analysis favoring pembrolizumab in Europe (HR 0.49).

Head and neck malignancies

KEYNOTE-048 evaluated pembrolizumab in recurrent or metastatic head and neck squamous cell carcinoma (HNSCC)36 and showed an improved mOS when compared with cetuximab chemotherapy (8.4 vs 6.9 months). In CheckMate 141, patients with platinum-resistant recurrent/metastatic HNSCC demonstrated an ORR of 13.3% with nivolumab versus 5.8% in the standard therapy.37 Subgroup analyses favored nivolumab in patients aged <65, male, white, and from North America. However, 83% of the patients were white and <4% were black. These trials led to the FDA approval of pembrolizumab and nivolumab for recurrent or metastatic HNSCC.

Thoracic malignancies

In non-small cell lung cancer (NSCLC), nivolumab showed improved PFS and OS (9.2 vs 6.0 months) over docetaxel in CheckMate 017.38 Improved OS was observed in patients <65 years old (HR 0.52), 65–75 years old (HR 0.56), men (HR 0.57), white race (HR 0.59), and from Europe (HR 0.50). KEYNOTE-042 demonstrated pembrolizumab’s superiority over chemotherapy in untreated NSCLC regardless of PD-L1 status, with 5-year OS of 16.4 versus 12.1 months (HR 0.79).39 40 Subgroup analysis performed at the first interim analysis showed superior OS with pembrolizumab across PD-L1 tumor proportion score (TPS). For TPS ≥50%, HRs was 0.58 for ≥65 years old, 0.68 for males, and 0.65 globally. PD-L1 inhibitors like durvalumab, atezolizumab, and avelumab also show efficacy in NSCLC. Other studies like CheckMate 057, CheckMate 227, KEYNOTE-010, KEYNOTE-024, CheckMate 026, KEYNOTE-189, KEYNOTE-407, IMpower150, and PACIFIC showed improved outcomes with ICB for NSCLC. However, <21% of patients were non-white, 68% enrolled outside the USA, and none reported analysis based on race.41–49

ICB has also been studied in extensive stage small cell lung cancer (SCLC). In IMpower133, atezolizumab demonstrated improved mOS (12.3 vs 10.3 months) and PFS (5.2 vs 4.3 months) with manageable AEs.50 Subgroup analyses favored atezolizumab, particularly in patients aged ≥65 years. Other trials like CASPIAN and KEYNOTE-604 reported OS benefits with durvalumab and pembrolizumab combinations, respectively, while studies with nivolumab/ipilimumab maintenance and adebrelimab showed varied outcomes.51 52 In CASPIAN, >80% of enrolled patients were white, 1% were black, and >50% were age ≤65 years. Subgroup analysis favored durvalumab in patients aged ≤65 years, men, non-Asians, and individuals from Europe. KEYNOTE-604 reported benefit regardless of geographical location but no race was reported. ASTRUM-005 mostly enrolled patients in Asia and revealed a significant OS benefit with serplulimab augmentation to platinum-based chemotherapy regardless of age.53 Race was not reported in the study.

Breast cancer

Early ICB monotherapy trials in breast cancer showed limited benefit. In IMpassion130, atezolizumab with nab-paclitaxel improved PFS but not OS.54 IMpower131 also failed to show OS benefit.55 In KEYNOTE-355, pembrolizumab with chemotherapy improved PFS and OS in triple-negative breast cancer (TNBC) with PD-L1 combined positive score ≥10.56 Neoadjuvant trials like KEYNOTE-522 demonstrated superior pathologic complete response rates and favorable EFS with pembrolizumab plus chemotherapy in early stage TNBC.57 Although these were international studies, the total representation of non-whites across the studies ranged from 28% to 42%. Of these studies, IMpassion130 and 131 reported subgroup racial analysis, with IMpassion130 showing improved PFS in whites (HR 0.78).

Hematology malignancies

The clinical use of immune checkpoint inhibitors has been more limited in hematologic malignancies, with notable exceptions in certain lymphomas like classical Hodgkin’s lymphoma (cHL), where they have represented a major addition to the treatment armamentarium. Like many trials for hematologic malignancies, these studies have typically involved modest numbers of participants and subgroup analyses of outcomes were often not published. Relevant information from these trials is summarized in table 2 and online supplemental table 2.

Table 2

Key phase 2 and 3 studies evaluating ICB in hematologic malignancies

Classical Hodgkin’s lymphoma

Among hematologic malignancies, cHL was among the first where programmed cell death protein-1 (PD-1) blocking antibodies found application, with numerous studies demonstrating remarkable efficacy regardless of line of therapy. This lymphoma is distinguished by frequent genetic alterations of chromosome 9p24.1, resulting in the overexpression of PD-L1 and PD-L2 in the malignant cells.58

The use of nivolumab and pembrolizumab in relapsed/refractory (r/r) cHL were tested in CheckMate 20559 60 and KEYNOTE-08761–63, respectively. Recently published 5-year follow-up data of CheckMate 205 demonstrated an ORR of 71%, with a CR rate of 21%, and an OS of 71% at a median follow-up of 58.5 months.60 The median age of participants was 34 years, 42% were female, and most were white (86%). No subgroup analyses were reported. Similar outcomes were observed in KEYNOTE-087, which evaluated pembrolizumab monotherapy in r/r cHL. The 5-year follow-up showed an ORR of 71% with a CR of 27%.61 Women were 46% of participants and the median age was 35 years old. Racial and ethnic demographics were not reported.

KEYNOTE-204 showed a significant mPFS benefit for pembrolizumab versus brentuximab vedotin (BV) monotherapy in r/r cHL (13.2 vs 8.3 months).64 Women constituted 43% of the participants. Subgroup analysis by sex demonstrated an mPFS benefit for pembrolizumab compared with BV in women (15.2 vs 5.7 months, HR 0.49) but was not significant in men (11.4 vs 8.4 months, HR 0.75). The median age of participants was 36 years old, with those aged ≥65 years old constituting 16% of the participants. Patients <65 years old had better mPFS with pembrolizumab (19.3 vs 8.3 months, HR 0.59), whereas no significance was observed for those ≥65 years old (8.2 vs 5.5 months, HR 0.64). Race and ethnicity were not reported.

The SWOG S1826 trial evaluated the efficacy of nivolumab with doxorubicin, vinblastine, and dacarbazine (AVD) versus BV+AVD among patients aged ≥12 years old with untreated advanced-stage cHL.65 This study was halted early following interim analysis that revealed nivolumab-AVD met the protocol-specified efficacy bounds compared with BV-AVD. A total of 994 patients were enrolled across the USA, with a median age of 27 years (10%>60, 24%<18), 44% female, and 76% white, 12% black, and 13% Hispanic. Subgroup analysis of patients >60 years old also demonstrated benefit for nivolumab-AVD versus BV-AVD.66 Notably, non-relapse mortality was 4% for nivolumab-AVD versus 14% for BV-AVD.

Primary mediastinal B-cell lymphoma

CheckMate 436 compared nivolumab versus BV in patients with r/r primary mediastinal B-cell lymphoma (PMBCL).67 68 This study had 29 evaluable patients and found an ORR of 73%, with 24 months PFS and OS rates of 56% and 76%, respectively. Median age was 36 years old and 56% of participants were women. KEYNOTE-170 evaluated pembrolizumab monotherapy in 53 patients with r/r PMBCL who were ineligible for autologous stem cell transplantation (ASCT) and reported an ORR of 45% with CR rate of 13%.69 Collectively, women made up 59% of participants and median age was 33 years old. No racial or ethnic information and no subgroup analysis was published for either study.

Other B-cell malignancies

Evidence for the efficacy of PD-1 blockade in other B-cell malignancies is limited. CheckMate 139 evaluated nivolumab in patients with r/r diffuse large B cell lymphoma (DLBCL) who had progressed after ASCT or in ASCT-ineligible patients.70 This study enrolled 121 patients and found an ORR of 10% in patients who had previously received ASCT, and 3% in ASCT-ineligible patients. Similarly, a single-arm phase 2 study of pembrolizumab maintenance after ASCT failed to show efficacy.71 Trials in r/r follicular lymphoma and other low-grade B-cell malignancies like chronic lymphocytic leukemia have not reported significant clinical benefit, though there has been evidence of activity of ICB for patients with Richter transformation.72–75

Peripheral T-cell lymphomas

A multicenter study evaluated monotherapy with geptanolimab, a PD-1-directed monoclonal antibody, in patients with r/r peripheral T cell lymphoma (PTCL) in 41 sites across China.76 Among 89 evaluable participants, ORR was 40% and CR was 15% but relatively high rates of AEs were observed. Women made up 31% of participants. The median age was 52 years old with 15% being >65 years old. No subgroup analysis based on age or sex were reported.

Two small studies have examined the use of pembrolizumab and nivolumab monotherapy in the USA for patients with PTCL. A phase 2 study evaluated nivolumab in 12 patients with advanced r/r PTCL (angioimmunoplastic T cell lymphoma and PTCL-not otherwise specified) and showed an mPFS of 2.7 months and OS of 6.7 months. However, hyper-progression of disease was seen in one-third of patients and the study was halted.77 Another phase 2 multicenter study evaluated pembrolizumab after ASCT for patients with PTCL in first remission.78 In total, 21 patients were enrolled, and 83.6% met the primary outcome of 18-month PFS. OS was 94.4%, which was better than the natural history post-ASCT. One-third of patients had immune-related AEs and four patients discontinued the study due to toxicity. In this study, 43% were women and the median age was 58 years old. White made up 52%, blacks were 10%, and Asian patients were 10%.

Other T-cell lymphomas

While there is evidence of response to ICB in natural killer/T-cell lymphoma, there are no prospective trials specifically in this population. Two small retrospective studies conducted in Asia have examined the efficacy of pembrolizumab and nivolumab showing high response rates including complete responses (CRs).79 80 A phase 2 trial of pembrolizumab in r/r advanced staged mycosis fungoides and Sezary syndrome enrolled 24 patients and demonstrated an ORR of 38% and CR of 8%.81 Responders tended to have durable response (median duration of response not reached at 1 year). The study enrolled of six women (25%) and the median age was 67 years old. Racial/ethnic demographics were not reported. Immune-related AEs were seen in 11 (45%) patients, 4 (17%) had to discontinue therapy.

Multiple myeloma

There had been significant interest in the use of ICB in multiple myeloma (MM), but studies demonstrated an unfavorable safety profile in this disease, leading to regulators halting the largest studies. Interim analysis of KEYNOTE-185 and KEYNOTE-183 found increased toxicity, including increased mortality, in the pembrolizumab arm of both trials.82 83 A total of 550 patients with r/r MM were enrolled in these studies, which showed no benefit and instead showed a 15% increased incidence of serious AEs over the control arms. There were 10 deaths in the pembrolizumab arm compared with 2 deaths in the control arm. In KEYNOTE-185, 54% of participants were women and the median age was 74 years old (21% ≥80 years old). Race/ethnicity demographics reported in ClinicalTrials.gov, showed participants consisted of 76% white, 4% black, 17% Asian, and 3% Hispanic. In KEYNOTE-183, 38% of participants were women, median age was 66 years old (7% ≥80 years old). Race/ethnicity demographics reported on ClinicalTrials.gov included 73% white, 10% black, 12% Asian, and 5% Hispanic. Subgroup analysis of outcomes were not reported.

Myeloid malignancies

A recent study conducted through the National Cancer Institute Cancer Therapy Evaluation Program enrolled 79 evaluable patients with acute myeloid leukemia (AML) who were not transplant eligible in the first remission and randomized to nivolumab maintenance versus observation.84 Overall, 2-year PFS, the primary endpoint, was similar in both arms (30%) but grade 3 or higher AEs were higher with nivolumab (71%) compared with observation (12%). Enrolled participants were 47% women, 70% white, 10% Hispanic, 9% black, and 3% Asian. Most were ≥60 years old (81%). No subgroup analysis was reported. Nivolumab in combination with azacitadine was studied 70 patients with r/r AML as an open-label phase 2 study. This study found an ORR of 33%, with higher responses observed in the hypomethylating agent-naïve patients (ORR 58%). The median age of participants was 70 years old. Sex and racial/ethnicity data was not reported. Early studies examining ICB therapy as maintenance after allogeneic stem cell transplant has been limited to due concerns for graft-vs-host disease and immune-related AEs.85

Retrospective and population-based studies

Several groups have also used retrospective and population-based approaches to examine the effect of race, ethnicity, sex, age and social determinants of health on the efficacy and safety of ICB therapy:

Retrospective studies that explored the impact of race and ethnicity have mostly focused on NSCLC and have reported mixed findings. A study of 11,138 patients treated with anti-PD-1/PD-L1 therapy reported that blacks had similar response rates but longer OS compared with whites.86 Another study evaluated 186 patients and found that Hispanics had inferior PFS and OS but also reduced AEs.87 Inferior outcomes for Hispanics and blacks compared with whites was also reported in a single-center study of 207 patients.88 A population-based claims study using the Surveillance, Epidemiology, and End Results (SEER)–Medicare database did not find a difference in 2-year OS based on race/ethnicity among patients who received ICB for NSCLC (30% for white, 33% Asian, 26% for black, and 28% for Hispanic).89 However, only 29% of blacks received ICB compared with 40% of white patients.

Clinical studies examining sex-based differences in efficacy outcomes have also shown mixed findings. A meta-analysis of the 39 randomized clinical trials found that both men and women benefited from ICB therapy, with no apparent differences based on sex.90 An SEER-Medicare evaluation of 1,369 patients also found no sex-based differences in outcomes among patients receiving anti-PD-1 therapy but combination therapy after ipilimumab showed a higher mortality hazard than anti-PD-1 therapy only in women.91 Studies are more consistent in reporting increased AEs in women,92 prompting the design of a prospective observational study (G-DEFINER, NCT04435964) specifically aimed at better understanding the association of sex and AEs.93 Prospective studies like this one will help determine the magnitude and the mechanistic basis for any differences that could be explained by sex.

The social determinants of health such as income, insurance status, distance from care, social support, or living in urban/rural setting significantly impact cancer therapy outcomes. However, these variables are almost never reported in clinical trial publications and very few studies have explored the association of these variables with outcomes after ICB. While social determinants adversely impact the stage at presentation and access and utilization of ICB,89 a single-institution study reported similar outcomes after ICB regardless of the patients’ insurance status, employment status, or neighborhood poverty level.94 Larger, multi-institutional studies are needed to more adequately assess the effects of these variables on treatment outcomes.

Discussion

In the decade since FDA approval of the first ICB therapy, there has been an explosion of clinical trials examining new indications for these drugs. Clinical benefits were observed in many cancers, with meaningful improvements in PFS and OS over prior treatment standards. As ICB becomes ubiquitously used, it is important to appreciate how representative these studies are of patients seen in clinical practice. Why does representation matter? Beyond biological reasons for the difference in outcomes, the benefits of having trials that represent the treated population are multifold and include improving patient and clinician perceptions of legitimacy and confidence of trials results, improved generalizability, and ensuring equity and fairness in participant selection for the research enterprise. In the United States, there are concerted efforts to improve representation that have led to significant improvements in the representation of women in clinical research. However, progress is still lagging on the inclusion of racial and ethnic minority populations.95

Biological or social differences may indeed modify the efficacy and AE profile of a treatment. For example, prior studies have suggested differential efficacy and higher rates of AEs among women.91–93 96 Others have found differential outcomes by race and ethnicity,87 88 but age does not seem to reduce the efficacy of ICB.97 98 These differential outcomes, however, may not be solely driven by biological differences, as race and ethnicity are social constructs99 that are typically reported based on the categories recommended by the US Office of Management and Budget. These broad categorizations may be proxies for potentially biologically relevant variables such as genetic ancestry and human leukocyte antigen (HLA) subtypes which could plausibly modify the biological response to ICBs. They may also be proxies for social variables such as environmental exposures, socioeconomic status, and access to care. While retrospective studies can provide insights on outcomes of ICB in subgroups that are under-represented in investigational trials, these studies may be confounded by institutional practice patterns, selection bias, and limitations in using clinical data which may not be well suited for these types of studies. This may explain the mixed findings reported thus far. Prospective studies that evaluate outcomes and AEs with measures of biological correlates will be essential to determine the differences in efficacy and AEs for ICB for biological and social variables.

What does our review tell us? First, while virtually all trials reported the sex and some metric for the age of participants, many did not report race/ethnicity. These data were sometimes reported to ClinicalTrials.gov but reporting and categorization was not consistent. Other biologically relevant variables like obesity or body mass index were rarely reported. Second, subgroup analyses for the primary endpoints were conducted in many studies within solid malignancies, but rarely in hematologic malignancies. This may be partly due to fewer participants in the hematological malignancy studies making subgroup analysis statistically infeasible. When subgroup analysis was conducted, typically a qualitative benefit of ICB were seen across all subgroups, though the magnitude of efficacy may be diminished for some, or more often, the effect size may be similar but statistical significance was not reached due to lack of statistical power. Third, little to no emphasis was placed on conducting subgroup analyses of AEs. Lastly, the trials examined were predominantly enrolling participants in North America, Europe, and East Asia; broad swaths of the global south did not have significant participation in these studies.

Several barriers to enrolling a diverse population in clinical trials have been identified. These include geographic restrictions such as difficulty accessing institutions offering trials and sparsity of trials near under-represented communities, lack of awareness, limited social and economic support, language barriers, and mistrust in the healthcare system. Trial-specific criteria such as baseline laboratory values can also contribute to disparities in the study population.100 In the studies we reviewed, women appear to be relatively appropriately represented and age and sex demographics are largely expected for a given cancer. Asian representation was high in some studies but that was only in international trials where patients were able to enroll in Asia. Indeed, most studies based in the USA under-represented black, Asian, and Hispanic participants. Renewed efforts are needed to offer greater access to ICB trials nationally and internationally and to gain the trust of traditionally marginalized populations. This can be done through decentralizing clinical trials, increasing the diversity in the healthcare workforce, creating language-concordant educational material, and design trials that appropriately support participants and minimize the number of trial-related visits and procedures. The work of the US national cooperative groups are examples of how increasing access and education result in more representative study populations. These studies, which enroll at numerous sites, have had greater success at enrolling a diverse participant pool compared with studies that accrue at few academic centers. In cHL, for example, the CheckMate 205 (38 locations in largely academic centers) had a similar proportion of male and female participants but significantly fewer racial and ethnic minority groups compared with the SWOG S1826 trial (736 locations), which more closely approximated the demographics of the USA.

In conclusion, despite the rapid adoption of ICBs for the treatment of many cancers, our understanding of the impact of race, ethnicity, sex, age and social determinants on outcomes following ICB remains limited. Unfortunately, racial and ethnic minorities remain largely under-represented in ICB trials and inconsistencies in how studies report such demographic information further limits the possibility of performing additional analyses. Addressing barriers that limit the enrollment of under-represented individuals and implementing standards for reporting patient demographics and outcomes constitute the initial step for our improved understanding of how these factors impact clinical outcomes. Ultimately, ensuring that there is a diverse representation in research and healthcare is essential to bettering treatment for all.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

Not applicable.

Acknowledgments

This work was partly supported by the institutional funds from Nell W and William S Elkin Fellowship in Oncology (RKN), Winship Cancer Institute of Emory University Winship Invests award, the NIH 1K08AI178093, the CLL Society Young Investigator Award, and the American Society of Hematology Scholar Award to AC, the NIH T32CA160040 for which AC is an awardee, and the K12CA237806 from the Emory K12 Clinical Oncology Training Program for which AC and ZSB are awardees. Support was also provided by NIH U54CA274513, and K08CA270401, and the Department of Defense grant BC2207P1 to ZSB. Additional support was provided by the Georgia Clinical & Translational Science Alliance under NIH Award UL1TR002378. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.