Real-World Immune-Related Adverse Events in Patients with Early Triple-Negative Breast Cancer Who Received Pembrolizumab

Abstract

PURPOSE:

The addition of pembrolizumab to chemotherapy in high-risk early triple negative breast cancer (TNBC) improves cancer outcomes. However, pembrolizumab induces varied immune-related adverse events (irAEs) where some can be severe or life-long. This retrospective study describes real-world patterns of irAEs in patients with TNBC who received pembrolizumab.

METHODS:

We evaluated irAEs in patients with TNBC from a comprehensive cancer center and a community hospital who received pembrolizumab with chemotherapy between 2021 and 2023, excluding those enrolled in clinical trials. We used national guidelines to grade toxicities. Logistic regression assessed the effect of clinicopathological variables on irAEs adjusting for covariates.

RESULTS:

We identified 233 patients with a median age of 51 years, 62% had stage II, 35% had stage III TNBC, 25% were Hispanic, 21% were Black and 42% were White. Eighty patients (34%) developed 100 separate irAEs. The most common irAEs were endocrinopathies (52%) and gastrointestinal (23%); there were 26 grade≥ 3 irAE which all resulted in hospitalization, the most common being gastrointestinal (13 instances); 45 required systemic steroids, 16 required additional immunosuppressive therapy and 32 patients discontinued pembrolizumab due to irAEs. Two patients who developed colitis eventually died due to complications. Most (67 instances) irAEs were unresolved at the time of last follow up, but 55% (37/67) had improved to grade 1. No clinicopathologic factors were associated with the development or severity of irAEs.

CONCLUSION:

In this real-world diverse population we observed rates of irAEs comparable to KEYNOTE-522, where endocrinopathies were the most prevalent, but gastrointestinal irAEs were also prevalent and severe. This emphasizes a critical issue as pembrolizumab is increasingly being used in early TNBC and could have long-term survivorship implications.

INTRODUCTION

Triple negative breast cancer (TNBC) represents approximately 15% of invasive breast cancers and is defined as the absence of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor 2 (HER2) expression.¹ Compared with other breast cancer subtypes, TNBC is more aggressive and has a greater likelihood for early relapses and poorer survival.^2,3 While its aggressive tumor biology is one of several factors that can contribute to racial disparities in breast cancer mortality, there are other non-biologic factors mediating this effect.^1,4 For example, TNBC disproportionally affects African American women.⁴

Chemotherapy in combination with the immune checkpoint inhibitor (ICI) pembrolizumab is the first line standard of care in metastatic TNBC positive for Programmed Cell Death Ligand 1.⁵ The addition of pembrolizumab to neoadjuvant chemotherapy followed by adjuvant pembrolizumab was approved in high-risk (clinical stage II-III) TNBC based on the KEYNOTE-522 phase 3 clinical trial.⁶ This study revealed a significantly improved pathologic complete response rate (64.8% vs 51.2%) with pembrolizumab and it shifted pateints into lower residual cancer burden categories. Overall, 15.7% of pateints who received pembrolizumab plus chemotherapy suffered an event (7.7% were distant recurrences) compared to 23.8% among those who only received chemotherapy (13.1% distant recurrences).⁷

While the incorporation of pembrolizumab to chemotherapy has improved cancer outcomes in TNBC, the development of immune-related adverse events (irAEs) can contribute to substantial morbidity as these are autoimmune conditions that can affect a myriad of organs or systems resulting in a wide range of clinical presentations that can occur any time during pembrolizumab administration and even after discontinuation.⁸ Interestingly, exploratory analyses show a positive association between patients who developed an irAE and an anti-tumor response to ICI.^8,9. The American Society of Clinical Oncology (ASCO) irAE guidelines and Common Terminology Criteria for Adverse Events (CTCAE v5) guidelines are useful tools for identifying the various presentations of irAEs by organ or system along with the severity to standardize grading of these toxicites.^10,11

In KEYNOTE-522 the reported rate of all grade irAEs was 33.5%. The most common system involved was endocrine, with hypothyroidism (15.1%), hyperthyroidism (5.2%), and adrenal insufficiency (2.6%) being the most frequent. Severe skin reactions occurred in 5.7% of patients who received pembrolizumab. Overall, 12.9% of patients who received pembrolizumab experienced grade ≥3 irAEs, with severe skin reactions (4.7%) being the most common followed by endocrinopathies.^6,12 However, the rate, severity and clinical characteristics of irAEs in real-world patient populations are not well described.

Although irAEs have a broad heterogenous presentation, the treatment of an irAE depends on the severity but often involves holding immunotherapy and administering systemic steroids or other forms of immunosuppression (e.g. anti-TNF infliximab in steroid refractory cases).¹³ While treatment guidelines are helpful, irAEs may result in treatment delays, life-long toxicities, and even death. There is no data available regarding any clinicopathological factors that may correlate with a higher risk of developing irAEs. Understanding these predictors is a key research priority to mitigate treatment toxicity.

This retrospective study analyzes real-world patterns of irAEs in patients with TNBC who received pembrolizumab with chemotherapy at The University of Texas MD Anderson Cancer Center (MDACC) and the community county hospital Lyndon B. Johnson (LBJ) which treats a high proportion of underserved and minority populations. This study is important for clinical practice because it includes an underrepresented population and it provides a real-world experience of the toxicities associated to the introduction of pembrolizumab into the standard treatment of early TNBC, which is becoming a more common practice in the onocology community and it could have life-long consequences for some patients.

METHODS

The objectives of this study were to describe the rate and severity (grade) of irAEs, to identify clinicopathological characteristics that could be associated with the development of irAEs, and to describe the natural history of irAEs in a real-world diverse patient population with early TNBC who received pembrolizumab.

Patient Population

We retrospectively reviewed the electronic medical records of adult patients (age 18 years or older) diagnosed between 2021 and 2023 with early TNBC (ER and PR ≤ 10% by immunohistochemistry [IHC] and HER2-negative [IHC-0, IHC-1+, or IHC-2+ with negative FISH]) who received neoadjuvant pembrolizumab in combination with chemotherapy at MDACC or at LBJ. We excluded patients with metastatic disease, incomplete medical documentation, or those who participated in therapeutic clinical trials.

Patients from MDACC were identified through the Department of Breast Medical Oncology Breast Cancer Management System, which is a large, well-annotated prospectively gathered database of patients with breast cancer evaluated at MDACC. Follow-up information is obtained via scheduled and routine review of the electronic medical records and also through a direct link to the Tumor Cancer Registry which queries the Social Security Death Index and the Texas Bureau of Vital Statistics to determine the vital status. Patients from LBJ were identified through the EPIC electronic health record SlicerDicer tool, which identified patients with TNBC who received neoadjuvant pembrolizumab plus chemotherapy using the Beacon Oncology module. This study was approved by each hospital’s Institutional Review Board.

Data Collection Procedures

We extracted demographic, clinicopathological and treatment-related data available in the database and by review of the electronic medical records. Demographic and clinicopathological data included race/ethnicity, gender, age at breast cancer diagnosis, histopathology, tumor grade, Ki-67, clinical tumor stage, types of neoadjuvant regimens received and Body Mass Index (BMI) at first pembrolizumab dose.

A treatment toxicity was considered an irAE if it met one or more of the following criteria: pathologic confirmation if a biopsy was performed, clinical improvement with treatments for irAE (e.g. immunossupressive therapy) or a documented diagnosis of irAE by a physician. We extracted the following information of each irAE: type, attribution, date of onset, use of systemic steroids and/or other systemic immunosuppressive therapy, hospitalizations due to irAE, resolution (if applicable), permanent discontinuation of pembrolizumab due to irAE, and associated deaths. We used both the ASCO irAE guidelines¹⁰ and the CTCAE v5 guidelines¹¹ to grade each irAE and we recorded the peak toxicity grade. If there was a discrepancy of grades between both guidelines, we documented the highest toxicity grade. We defined an unresolved irAE as one where the patient was still symptomatic or requiring treatment for such toxicity at the last follow up clinical visit. We obtained data on the irAE timing and duration, including the following time points: date of first pembrolizumab treatment, date that the irAE was identified, and date the toxicity was resolved (if applicable), defined as a toxicity grade of 0 without the requirement for medications.

Statistical Analysis

Data were summarized using standard descriptive statistics such as mean, standard deviation, median, and range for continuous variables, and frequency and proportion for categorial variables. The Kaplan-Meier method was used to estimate probability of irAE free survival. The association between categorical variables was examined by Fisher’s exact test. The Wilcoxon rank sum test was used to examine the difference in continuous variables between two groups. All computations were carried out in SAS 9.4 (SAS Institute Inc., Car, NC, USA) and TIBCO Spotfire S+.

RESULTS

We identified 233 female patients who met the criteria for the study with a median age of 51 ± 12 years; 62% had stage II and 35% had stage III TNBC. Regarding race/ethnicity, 25% were Hispanic/Latino, 21% were non-Hispanic Black, 6% were Asian and 42% were non-Hispanic White. The most common histology was invasive ductal carcinoma (94%) followed by other histologies (4%) which included metaplastic carcinoma and invasive carcinoma with neuroendocrine features; very few were invasive lobular carcinoma (1.3%) or invasive mixed ductal/lobular (0.4%). The average BMI at first pembrolizumab dose was 29.68 ± 6.54. The breakdown of patient characteristics is displayed in Table 1. Patients received four different types of pembrolizumab-containing systemic therapy regimens including the KEYNOTE-552 regimen of paclitaxel plus carboplatin followed by doxorubicin and cyclophosphamide (AC) (79%), taxane plus carboplatin (19.3%), AC (1.3%), and taxane plus AC (0.4%).

Table 1.

Patient clinicopathologic variables

Variables	All Patients (n = 233)	MDACC (n = 198)	LBJ (n = 35)
Race/ethnicity, No. (%)
Asian	15 (6.4)	15 (7.6)	0
Hispanic/Latino	57 (24.5)	34 (17.2)	23 (65.7)
Non-Hispanic Black	48 (20.6)	40 (20.2)	8 (22.9)
Non-Hispanic White	97 (41.6)	95 (48)	2 (5.7)
Other	16 (6.9)	14 (7.1)	2 (5.7)
Age at diagnosis, years, mean ± SD, median (min, max)	51 ± 12, 50 (23, 80)	51 ± 12, 51 (23, 78)	50 ± 12, 48 (29, 80)
BMI, mean ± SD, median (min, max)	29.68 ± 6.6, 28.22 (18.96, 49.69)	29.16 ± 6.04, 27.75 (18.96, 47.99)	32.58 ± 8.37, 30.22 (20.2, 49.69)
Ki-67, %, mean ± SD, median (min, max)	70.76 ± 22.51, 75 (5, 100)	69.08 ± 22.56, 72.5 (5, 100)	78.86 ± 20.73, 90 (16, 99)
Tumor grade, No. (%)
I	3 (1.3)	3 (1.5)	0
II	27 (11.5)	25 (12.6)	2 (5.7)
II	203 (89.7)	170 (89.4)	33 (94.3)
Clinical T, No. (%)
T0	4 (1.7)	3 (1.5)	1 (2.9)
T1	23 (9.9)	21 (10.6)	2 (5.7)
T2	129 (55.4)	116 (58.6)	13 (37.1)
T3	48 (20.6)	42 (21.2)	6 (17.1)
T4	29 (12.4)	16 (8.1)	13 (37.1)
Clinical N, No. (%)
N0	116 (49.8)	104 (52.5)	12 (34.3)
N1	69 (29.6)	59 (29.8)	10 (28.6)
N2	13 (5.6)	10 (5.1)	3 (8.6)
N3	35 (15)	25 (12.6)	10 (28.6)
Clinical stage, No. (%)
I	8 (3.4)	8 (4)	0
II	144 (61.8)	134 (67.7)	10 (28.6)
III	81 (34.8)	56 (28.3)	25 (71.4)
Histopathology, No. (%)
Invasive ductal carcinoma	219 (93.9)	184 (92.9)	35 (100)
Invasive lobular carcinoma	3 (1.3)	3 (1.5)	0
Invasive mixed ductal and lobular	2 (0.8)	2 (1.0)	0
Other *	9 (3.9)	9 (4.5)	0

BMI, body mass index; LBJ, Lyndon B. Johnson Hospital; MDACC, The University of Texas MD Anderson Cancer Center; SD, standard deviation

^*Other: metaplastic, n= 4; invasive carcinoma with neuroendocrine features, n= 3; apocrine adenocarcinoma of breast, n= 1; adenoid cystic carcinoma, n= 1

Eighty patients (34%) developed 100 distinct irAEs (94 were definitely attributable to pembrolizumab and 6 were possibly or probably attributable), where 18 patients (8%) experienced more than one (multiorgan/multisystem) irAE. Figure 1 shows irAE-free survival of all 233 patients: at 30 days 98% were irAE-free, at 60 days 91%, at 120 days 82%, and at 180 days 75% were irAE-free. Table 2 provides a breakdown of the specific types of irAEs per organ or system along with the toxicity grade. Table 3 shows the median time in days for the development of the first irAE per organ or system among the 80 patients who developed at least one irAE. Figure 2 shows all reported irAEs by organ or system and by institution. The most common irAEs were endocrinopathies (52%), where hypothyroidism was the most common (32%) followed by adrenal insufficiency (15%). The second most common system affected was gastrointestinal (23%), where colitis and hepatitis were the most common (both 10%). Regarding the peak toxicity grade of all recorded irAEs, 30% were grade 1, 44% were grade 2, 20% were grade 3, 4% were grade 4, and 2% were grade 5 (two patients died after complications of colitis). Twenty-six grade ≥ 3 irAE occurred in 24 patients (10.3%) with the most common system being gastrointestinal, followed by adrenal insufficiency and pneumonitis. Two patients suffered 2 simultaneous grade ≥ 3 irAEs: one had a grade 3 endocrinopathy and grade 3 nervous system irAE, and another one had a grade 4 gastrointestinal and grade 3 hematologic irAE.

Figure 1. Immune-related adverse event-free survival.

Figure 1 shows the Kaplan-Meier estimates of irAE-free survival measured in days which represents the chance of being irAE-free in 233 patients with early TNBC who received pembrolizumab plus chemotherapy. Tick marks represent data censored at the last time that a patient was known to be alive and without an irAE of any grade. An irAE was defined as a toxicity derived from treatment with pembrolizumab which was documented as such by a physician, or had a pathological confirmation if a biopsy was performed, or there was clinical improvement of the toxicity after treatment with immunosuppressive therapy.

Table 2.

Immune-Related Adverse Events by Organ or System and Grade (N, %)

Organ or System	Any Grade (N= 100)	Grade ≥ 3 (N= 26)
Cutaneous	4 (4%)	0
Rash	4 (4%)	0
Endocrine	52 (52%)	5 (19%)
Hypothyroidism	32 (32%)	0
Adrenal insufficiency	15 (15%)	5 (19%)
Thyroiditis	5 (5%)	0
Gastrointestinal	23 (23%)	13 (50%)
Colitis	10 (10%)	7 (27%)
Hepatitis	12 (12%)	5 (19%)
Pancreatitis	1 (1%)	1 (4%)
Hematologic	4 (4%)	2 (8%)
Autoimmune hemolytic anemia	2 (2%)	2 (8%)
Thrombocytopenia	2 (2%)	0
Lung	6 (6%)	4 (15%)
Pneumonitis	6 (6%)	4 (15%)
Musculoskeletal	5 (5%)	0
Arthralgia	4 (4%)	0
Myositis	1 (1%)	0
Nervous system	4 (4%)	1 (4%)
Encephalitis	2 (2%)	1 (4%)
Aseptic meningitis	1 (1%)	0
Myasthenia	1 (1%)	0
Renal	1 (1%)	0
Nephritis	1 (1%)	0
Other	1 (1%)	1 (4%)
Anaphylaxis	1 (1%)	1 (1%)

A treatment toxicity was considered an irAE if it met one or more of the following criteria: pathologic confirmation if a biopsy was performed, clinical improvement with treatments for irAE or documented irAE by a physician. The ASCO irAE guidelines and the CTCAE v5 guidelines were used to grade the irAE using the highest applicable peak grade between both guidelines.

Table 3.

Median time in days to the first immune-related adverse event (N= 80 patients)

Organ or system	Median time (days) to irAE (Range)
All	117.5 (1–623)
Cutaneous (rash)	241 (126–312)
Endocrine a	125.5 (21–515)
Gastrointestinal b	63.5 (7–206)
Hematologic c	47 (1–147)
Lung (pneumonitis)	331 (40–369)
Musculoskeletal d	206 (89–283)
Nervous System e	338.5 (54–623)
Other (anaphylaxis)	167.5 (50–285)
Renal (nephritis)	77 (77–77)

^ahypothyroidism, adrenal insufficiency, thyroiditis

^bcolitis, hepatitis, transaminitis, pancreatitis

^cautoimmune hemolytic anemia, thrombocytopenia

^darthralgias, myositis

^eencephalitis, aseptic menigitis, myasthenia

Figure 2. Breakdown of all grade irAEs (n=100) by organ or system per institution.

Figure 2 shows the absolute number of all grade irAEs categorized by organ or system and by each institution. The red bar represents data from The University of Texas MD Anderson Cancer Center (MDACC) which is located below the blue bar which represents data from Lyndon B. Johnson (LBJ) hospital.

Twenty-six irAE instances resulted in hospitalization (11 due to gastrointestinal irAEs), 45 instances required systemic steroids and 16 of these required additional immunosuppressive therapy. Most (67 instances) irAEs were unresolved at the time of the last clinic follow up visit, but 55% (37/67) had improved to grade 1. Among patients who had unresolved irAEs, the average number of days from the onset of an irAE to last clinic follow up was 259 ± 182 days (median time of 271 days, range 1–754 days) while the average number of days from the onset of an irAE to resolution was 122 ± 125 days (median time of 81 days, range 1–457 days). Thirty-two patients (14%) discontinued pembrolizumab due to development of irAEs. Patients were followed for a median time of 437 days (range 62 – 858) from the time of the first dose of pembrolizumab.

In this real world diverse population of patients with TNBC, no clinicopathologic factors were found to be associated with the development of, or with the toxicity grade of pembrolizumab-induced irAEs. Of specific interest, the following clinicopathologic variables were not associated with the development of irAEs: race/ethnicity (p= 0.49), clinical tumor stage (p= 0.56) , tumor grade (p= 0.09), age at diagnosis (p= 0.73) and baseline BMI (p= 0.21).

DISCUSSION

This retrospective study of a real-world diverse patient population with TNBC who received pembrolizumab showed similar rates and severity of irAEs as compared to those reported by the KEYNOTE-522 trial. However, the breakdown of the irAEs by organ or system was different, and one unique aspect of our study was having a diverse patient population. The community county hospital LBJ treats a high proportion of underserved and minority patients which is more representative of a diverse race/ethnicity and of a diverse socioeconomic status. Specifically, our study included 25% of Hispanic/Latino and 21% non-Hispanic Black patients. Although the demographic variable of race/ethnicity was not reported in the KEYNOTE-522 study, 50% of the patients were from Europe, 21% from Asia and 21% from North America.

The KEYNOTE-522 trial reported that 33.5% of patients who received pembrolizumab developed an irAE of any grade, while 12.9% suffered grade ≥ 3 toxicities and 2 (0.3%) patients died.¹² Our cohort had 34% of patients develop an irAE of any grade, 10.3% suffered grade ≥ 3 irAEs and 1% (2 patients) died due to complications of pembrolizumab-induced colitis. Although an accurate comparison is limited due to differences in the methodology of collecting data and assigning attributions of toxicities, the overall prevalence of all grade and grade ≥ 3 irAEs were similar, but the breakdown of organs and systems involved was different.

Similar to KEYNOTE-522, endocrinopathies were the most prevalent irAEs in our cohort with a comparable breakdown of hypothyroidism, adrenal insufficiency and thyroiditis. The average time reported for the development of endocrine irAEs is 16 to 26 weeks.¹⁴ Our study reports a median time of around 18 weeks (Table 3). Different than non-endocrine irAEs that may resolve with cessation of immunotherapy and immunosuppression, endocrine irAEs are for the most part irreversible and may require long term homone management.^15,16 Thyroid dysfunction often resembles thyroiditis but progresses rapidly to hypothyroidism with weeks and is managed with conservative measures and thyroid hormone replacement.^17,18 Pituitary dysfunction can induce primary or secondary adrenal insufficiency which requires an evaluation of cortisol and adrenocorticotropic hormone levels along with biochemical panels, and a symptomatic patient may require corticoid replacement therapy for life.¹⁹

Severe skin reactions reported in KEYNOTE-522 ocurred in 5.7% of patients who received pembrolizumab, of which 4.7% were grade ≥3. Our study only found 4 instances of a grade 1–2 pembrolizumab associated rash. It is possible that skin related toxicities were underreported given the retrospective design of our study and that a treatment related rash could also be attributed to chemotherapy.

The gastrointestinal system was the second most common involved with irAEs in our study, having colitis and hepatitis as the most prevalent. Gastrointestinal irAEs were also severe: 50% of all grade ≥ 3 irAEs and 38% of all irAE related hospitalizations were due to gastrointestinal irAEs, primarily colitis. Patients who have persistent toxicity beyond 3 months due to colitis have an increased risk of developing colon adenoma polyps that require more intense colonoscopy surveillance.²⁰ We report 2 patients who died due to medical complications after developing colitis: one patient received appropriate and prompt treatment with mesalamine, methylprednisolone, prednisone and vedolizumab for about 3 months but eventually developed hypotension and ischemic colitis; the second case developed neutropenic fever and septic shock despite antibiotic therapy and steroids in the context of colitis; interestingly, both pateints did not have significant comorbidities. Our results suggest that in a real-world population of patients with TNBC who receive pembrolizumab there could be a high rate of gastrointestinal irAEs, some of which could be severe. A rechallenge with ICI after an initial colitis event can still be safely practiced with the current strategy of early and aggressive treatment for moderate to severe irAEs where higher response and remission rates have been reported.²¹ For colitis the recurrence rate can reach 35% after a rechanllenge with ICI, but several treatment options have been explored that can reduce the recurrence to 15% or lower with concurrent biological maintenance or fecal microbiota transplantation yielding sustainable long term efficacy.^22,23

We did not identify clinicopathological factors such as race/ethnicity, age at diagnosis, histopathology, clinical tumor stage nor tumor grade that were associated with the development of irAEs. While there is ongoing research to determine biomarkers that could help predict which patients are more likely to develop an irAE ²⁴, these are currently not established. The I-SPY study presented a clinical irAE tool called PRO-CTCAE/PROMIS designed to help predict the development of irAEs.²⁵ This clinical trial showed that a greater number of White patients developed hypothyroidism compared to non-White patients (13% vs. 3%, p<0.04), and pneumonitis was more common in patients with an age >50 years old (p<0.002), suggesting that some clinical factors could be associated with an increased risk of developing an irAE.

Our results also showed a high rate (67%) of unresolved toxicities at the date of the last follow up clinic visit, although follow up time was varied due to the retrospective design. While KEYNOTE-522 did not report unresolved toxicities, other studies reported these in different tumor types. A retrospective study of 114 patients with non-small cell lung cancer who received ICI therapy reported that 24% developed long term (>1 year) irAEs.²⁶ Another retrospective study of 387 patients with stage III and IV melanoma treated with ICI therapy reported that 43% developed chronic irAEs: 96% were grade 1–2 and 86% persisted until their last follow up.²⁷ Our results suggest that chronic unresolved irAEs could be prevalent in a real-world patient population. We need to better understand this phenomenom and develop long-term survivorship plans in patients with cancer who receive ICI therapies.

Limitations of our study include patient selection bias as the sample was restricted to those who were treated at MDACC or LBJ and excluded patients who received partial care at these institutions.²⁸ We also relied on clinical documentation from providers to identify irAEs which can can have limitations in data availability or underreporting. There is a limitation in comparing the irAEs rates reported in our retrospective study with KEYNOTE-522, which underwent a prospective structured collection of irAEs in a careful and systematic manner and therefore any comparison is considered exploratory. We speculate that our retrospective study may have underreported low grade irAEs, such as skin toxicities. A real-world patient experience of more than 200 patients with TNBC is not ideal, but the addition of pembrolizumab to chemotherapy was recently approved in 2021. Another study also assessed irAEs in a real-world patients but had a smaller patient population.²⁹ Importantly, our patient population included a diverse group of patients from a county hospital that predominantly attends underserved population.

In conclusion, while the utilization of pembrolizumab in early TNBC has improved clinical outcomes, it is vital for oncologists to continue investigating irAEs. Although our study found similar rates and severity of irAEs compared to the KEYNOTE-522 trial, we found that gastrointestinal irAEs had higher toxicity grades. Some irAEs may need to be monitored more closely and proactively requiring earlier collaboration with specialists and discussing with patients the risks and benefits of continuing pembrolizumab once they develop an irAE. We did not find any clinicopathologic characteristics that correlated with development of irAEs which is an area of research that is greatly needed. Future retrospective and prospective large multi-institutional studies with diverse patient populations are needed to determine factors in the development of irAEs, the natural history incluing long-term consequences and survivorship care plans for patients with TNBC treated with pembrolizumab. At MDACC we developed a strategic initiative to develop and disseminate clinical best practices to improve outcomes in patients with irAEs.³⁰

Context Summary.

Key Objective

What are the patterns of immune-related adverse events (irAEs) in a real-world diverse population of patients with early triple negative breast cancer (TNBC) who received pembrolizumab?

Knowledge Generated

The rates of irAEs were comparable to those reported by the landmark clinical trial KEYNOTE-522. However, the breakdown by affected organs and systems was different. Endocrine irAEs were the most prevalent and some of them will require lifelong hormonal treatment. Gastrointestinal irAEs were more prevalent than reported in KEYNOTE-522 and some were severe enough that required pembrolizumab to be stopped, while some patients needed immunosuppressive therapy, and some were hospitalized. None of the clinical or tumor characteristics were associated with a higher risk of developing an irAE, which makes it difficult to prevent these toxicities.

Relevance

These findings are critical as pembrolizumab is increasingly being used to treat early TNBC by community oncologists which could have long-term survivorship implications.

Data Availability Statement:

The data that support the findings of this study are available by request from the corresponding author.