Immune-related adverse events among patients with early-stage triple-negative breast cancer treated with pembrolizumab plus chemotherapy: real-world data from the Neo-Real/GBECAM 0123 study

Abstract

Background

Pembrolizumab combined with neoadjuvant chemotherapy is the standard of care for stage II-III triple-negative breast cancer (TNBC) based on the KEYNOTE-522 trial. However, 13 % of patients experienced immune-related adverse events (irAEs) of grade ≥3 in the trial. This study aims to describe patterns of irAEs in a real-world scenario during treatment with pembrolizumab for early-stage TNBC.

Methods

Patients treated with neoadjuvant pembrolizumab plus chemotherapy across ten Brazilian cancer centers were evaluated in the Neo-Real/GBECAM0123 study. This analysis focuses on irAE evaluation, including time to onset, management, and association between irAEs and pathological complete response (pCR).

Results

A total of 368 patients were included. Overall, 31 % of patients (n = 114) presented with any grade irAEs. Most of irAEs (72.8 %) occurred during the neoadjuvant phase while 28.1 % happened during the adjuvant period. The most frequent irAEs were endocrine (12.8 % of the entire cohort), cutaneous (7.6 %) and gastrointestinal (7.1 %). A total of 50 patients (13.6 %) experienced grade ≥3 irAEs, predominantly gastrointestinal (32 %). 58 patients (56 %) needed corticosteroids. Immunotherapy rechallenge was possible in 53 % of the cases; permanent discontinuation of pembrolizumab was necessary for 16 %. No significant association was observed between irAEs and clinic-pathologic features nor pCR status.

Conclusions

In this real-world analysis, we observed a similar incidence of irAEs as reported in the KEYNOTE-522 trial. Most patients experienced resolution of their irAEs, but some required permanent discontinuation of pembrolizumab. Additionally, there were lasting dysfunctions, particularly endocrine, demanding lifelong support. Careful monitoring and management of these events are essential.

Highlights

• •This is a large real-world cohort of patients treated according to the KEYNOTE-522 regimen (368 patients from 10 institutions).
• •31 % of patients (n = 114) presented with any grade immune-related adverse events (irAEs), and 13.6 % were grade≥3.
• •Endocrine irAEs (12.8 %) were the most frequent and often persistent.
• •No significant association was observed between irAEs and clinic-pathologic features nor pCR status.
• •18.1% of patients had irAEs during adjuvant pembrolizumab, raising concerns about its use, especially after pCR.

1. Introduction

Breast cancer is the most common malignancy among women worldwide, with triple-negative subtype accounting for approximately 15 % of cases [1]. Currently, the standard treatment for high-risk localized disease involves the combination of chemotherapy and immunotherapy, as outlined in the KEYNOTE-522 trial [2].

The KEYNOTE-522 was a phase III trial that evaluated the role of combining chemotherapy (carboplatin and paclitaxel, followed by doxorubicin and cyclophosphamide) with pembrolizumab, an anti-PD-1 (programmed cell death protein 1) agent, in patients with stage II and III triple-negative breast cancer (TNBC). The regimen consisted of 3-weekly pembrolizumab combined with chemotherapy for eight cycles in the neoadjuvant phase, followed by surgery, and then up to nine cycles of pembrolizumab in the adjuvant setting. The first interim analysis reported an absolute benefit of approximately 13 % (64.8 % vs. 51.2 %) in the pathologic complete response (pCR) rate with the addition of pembrolizumab to neoadjuvant therapy [3]. This gain later translated into an improvement in event-free survival (5-year EFS 81.3 % vs. 72.3 %), and overall survival (5-year OS 86.6 % vs. 81.7 %) [[4], [5], [6]].

Despite the benefits of adding immunotherapy to neoadjuvant chemotherapy, there is a variety of potential immune-related adverse events (irAEs) associated with immune checkpoint inhibitors [7]. A pooled analysis of nearly 9000 patients who received pembrolizumab across 31 clinical trials, in 19 different cancer types, reported an irAE incidence of 23 %, with 6.3 % being grade 3 or higher, and 0.2 % associated with drug-related mortality [8]. In the KEYNOTE-522 study, serious treatment-related adverse events occurred in 34 % of patients receiving pembrolizumab-chemotherapy compared to 20 % in the placebo-chemotherapy group. Immune-related adverse events of any grade occurred in 35 % of patients in the pembrolizumab group, with 13 % experiencing irAEs of grade 3 or higher, and two deaths due to irAEs (0.3 %) [4,6].

Real-world safety data on new therapies are crucial outside of clinical trials to better assess adverse event patterns and potential safety signals in the general population [9]. Most real-world safety data about pembrolizumab come from melanoma and non-small cell lung cancer, for which immunotherapy has been approved earlier. [10,11]. Since the FDA approved pembrolizumab for early-stage TNBC in 2021, real-world data on the incidence of irAEs with this treatment regimen for early-stage TNBC are still scarce.

In this context, the Neo-Real/GBECAM-0123 study is a multicenter collaborative effort to generate real-world data on the use of neoadjuvant pembrolizumab immunotherapy in high-risk early-stage TNBC [12]. The present report aims to describe the incidence and patterns of irAEs in real-world settings during pembrolizumab treatment for early-stage TNBC, including types of adverse events, time to onset and duration, management, and resolution.

2. Methods

The Neo-Real/GBECAM 0123 study is a real-world collaborative study evaluating patients with early-stage triple-negative breast cancer treated with neoadjuvant pembrolizumab plus chemotherapy (CT) between July 2020 and July 2024 across ten Brazilian cancer centers. Patients who received at least one cycle of neoadjuvant pembrolizumab and had available data on adverse events were included in this current analysis. Patients were included based on the physician's choice to recommend the KEYNOTE-522 regimen, and the management of irAEs was defined by protocol, in accordance with international guidelines available to all recruiting centers. Cases selected for rechallenge were determined at the discretion of the treating physician, based on the severity of toxicity and the patient's recovery.

Data were retrospectively collected from electronic medical charts and registered using a centralized REDCap case report form. Baseline characteristics were presented using descriptive statistics, with continuous variables described as median and range and categorical variables as absolute and relative frequencies. ASCO (American Society of Clinical Oncology) guidelines for irAEs and the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0 were used for toxicity grading. Fisher's exact test was used for the comparison of categorical variables, Student's T test for normally distributed continuous variables and Mann-Whitney test for skewed continuous variables. Logistic regression analyses were conducted to evaluate possible clinical predictors of irAEs. The following clinical variables were evaluated in the univariate logistic regression: age, histologic type, histologic grade, tumor stage, body mass index, type of AC (doxorubicin plus cyclophosphamide) regimen (dose-dense or every 3-week), menopausal status, and use of granulocyte colony stimulating factor. The irAE-free survival (time from first cycle of pembrolizumab to irAE occurrence) was assessed using the Kaplan-Meier method. Statistical significance was defined as a p-value less than 0.05. Statistical analysis was performed using Stata software, version 15.1 (StataCorp, Texas, USA).

Ethical approval was obtained from the Institutional Review Boards of each of the participating institutions. The study adhered to the ethical guidelines established by the Declaration of Helsinki. Written informed consent was obtained from participants who could be contacted, while a waiver of consent was granted by the Institutional Review Boards for deceased or untraceable participants, as the study only collected de-identified data.

3. Results

3.1. Patients’ characteristics

In total, 368 patients were included, with a median age of 43.7 years (range 22.5–87.9). Most patients were premenopausal (63.3 %), with 70.9 % diagnosed with stage II breast cancer and 25.5 % with stage III breast cancer. Baseline characteristics are detailed in Table 1. A comparison of clinical and pathological features between patients with and without irAEs did not show any statistically significant difference (Table 1). At time of analysis, a total of 354 (96.2 %) patients completed the neoadjuvant phase of treatment and 284 entered the adjuvant phase of the protocol, of whom 177 (48.1 %) patients concluded the adjuvant phase (Supplemental Fig. 1).

Table 1.

Patients’ characteristics according to the incidence of immune-related adverse events (N = 368).

Patients' Characteristics	All patients (N = 368)	With irAE (N = 114)	No irAE (N = 254)	P-value
Age at diagnosis (median, range)	43.7 (22.5–87.9)	44.4 (23.9–78.5)	43.7 (22.5–87.9)	0.888a
Age≥65 years-old	36 (9.8 %)	12 (10.5 %)	24 (9.4 %)	0.850b
Menopausal status				0.164b
Premenopause	233 (63.32 %)	73 (64.03 %)	160 (62.99 %)
Postmenopause	110 (29.89 %)	29 (25.43 %)	81 (31.88 %)
Unknown	25 (6.79 %)	12 (10.52 %)	13 (5.11 %)
ECOG-PS				0.278b
0	326 (88.58 %)	97 (85.08 %)	229 (90.15 %)
1	22 (5.9 %)	8 (7.01 %)	14 (5.51 %)
2	1 (0.2 %)	1 (0.8 %)	0
Unknown	19 (5.16 %)	8 (7.01 %)	11 (4.33 %)
BMI (median, range)	26.15 (16.9–47.2)	25.9 (19.4–47.2)	26.3 (16.9–29.7)	0.882a
BMI>30 kg/m2	89 (24.2 %)	28 (24.6 %)	61 (24.0 %)	0.896b
Histology				0.714b
Ductal	328 (89.13 %)	102 (89.47 %)	226 (88.97 %)
Lobular	6 (1.63 %)	3 (2.63 %)	3 (1.18 %)
Other	18 (4.89 %)	5 (4.38 %)	13 (5.11 %)
Unknown	16 (4.35 %)	4 (3.50 %)	12 (4.72 %)
Grade				0.272b
I	3 (0.82 %)	1 (0.87 %)	2 (0.78 %)
II	60 (16.30 %)	18 (15.78 %)	42 (16.53 %)
III	266 (72.28 %)	79 (69.29 %)	187 (73.62 %)
Unknown	39 (10.6 %)	14 (12.28 %)	25 (9.84 %)
Ki67 (median, range)	70 (0–95)	80 (6–95)	70 (0–95)	0.105a
Clinical Stage				0.838b
I	9 (2.45 %)	2 (1.75 %)	7 (2.75 %)
II	261 (70.92 %)	85 (74.56 %)	176 (69.29 %)
III	94 (25.54 %)	26 (22.80 %)	68 (26.77 %)
Unknown	4 (1.09 %)	1 (0.87 %)	3 (1.18 %)
gBRCA1/2 mutation				0.474b
Yes	45 (12.23 %)	17 (14.91 %)	28 (11.02 %)
No	234 (63.59 %)	73 (64.03 %)	161 (63.38 %)
Unknown	89 (24.18 %)	24 (21.05 %)	65 (25.59 %)
Number of cycles of pembrolizumab before irAE (median, range)		5.0 (1–17)
Mean Total Number of Pembrolizumab Cycles		10.1	12.6
AC regimen				0.329b
Every 3 weeks	164 (41.84 %)	43 (37.71 %)	121 (47.63 %)
Every 2 weeks (dose-dense)	207 (56.25 %)	69 (60.52 %)	138 (54.33 %)
Other	6 (1.63 %)	2 (1.75)	4 (1.57 %)
Adjuvant therapy				0.852b
Olaparib + pembrolizumab	11 (2.98 %)	3 (2.6 %)	8 (3.15 %)
Capecitabine + pembrolizumab	78 (21.19 %)	23 (20.17 %)	55 (21.65 %)
Only pembrolizumab	214 (58.15 %)	56 (49.12 %)	158 (62.20 %)

AC: doxorubicin + cyclophosphamide. BMI: body mass index; ECOG-PS: Eastern Cooperative Oncology Group Performance Status. IrAE: immune-related adverse events.

^aWilcoxon rank-sum.

^bFisher's exact Test.

3.2. Incidence and types of irAEs

Overall, 114 patients (31 %) presented with any grade irAEs. Most of these patients (N = 83, 72.8 %) developed irAEs during the neoadjuvant phase, while 28.1 % (N = 32) happened during the adjuvant period. One patient had irAEs both during the neoadjuvant and the adjuvant phase. The median number of cycles of pembrolizumab to onset of irAEs was 5 (range 1–17). No baseline characteristics were associated with increased risk of irAE in the logistic regression (Supplemental Table 1).

A summary of all types of irAEs is represented on Table 2. The most frequent irAEs were endocrine (47 patients - 12.8 % of the entire cohort), cutaneous (28 patients - 7.6 %) and gastrointestinal (26 patients - 7.1 %), followed by pulmonary (15 patients - 4.1 %) and musculoskeletal (9 patients - 2.4 %) (Fig. 1A). Among the patients with endocrine irAE, there were 34 cases of thyroiditis (9.2 %), 7 adrenal insufficiencies (1.9 %), 5 hypophysitis (1.4 %) and 1 diabetes mellitus (0.3 %). The gastrointestinal events were represented by 16 patients with hepatitis (4.3 %), 9 colitis (2.4 %) and 1 pancreatitis (0.3 %). A total of 50 patients (13.6 % of the entire cohort) experienced grade ≥3 irAEs, predominantly gastrointestinal (17 patients - 4.6 %), with a 3.2 % incidence of grade ≥3 hepatitis and 1.4 % of grade ≥3 colitis, followed by endocrine (11 patients – 2.9 %), cutaneous (8 patients – 2.2 %), and pulmonary (7 patients - 2 %) (Fig. 2). No fatal irAEs occurred.

Table 2.

Types of irAEs in the whole cohort (N = 368).

irAE (N, %)	All grade	Grade≥3
Thyroiditis	34 (9.2)	5 (1.4)
Cutaneous	28 (7.6)	8 (2.2)
Hepatitis	16 (4.3)	12 (3.3)
Pneumonitis	15 (4.1)	7 (1.9)
Colitis	9 (2.4)	5 (1.4)
Arthritis	9 (2.4)	2 (0.5)
Primary adrenal insufficiency	7 (1.9)	3 (0.8)
Hypophysitis	5 (1.4)	2 (0.5)
Nephritis	3 (0.8)	2 (0.5)
Pancytopenia	2 (0.5)	2 (0.5)
Pure red cell aplasia	1 (0.3)	1 (0.3)
Amaurosis	1 (0.3)	1 (0.3)
Diabetes-mellitus	1 (0.3)	1 (0.3)
Cystitis	1 (0.3)	1 (0.3)
Hemophagocytic syndrome	1 (0.3)	1 (0.3)
Mononucleosis-like syndrome	1 (0.3)	1 (0.3)
Pancreatitis	1 (0.3)	0

irAE: immune-related adverse events.

Fig. 1.

A: Types of immune-related adverse events (all-grade) by system. B: Frequency of irAEs by phase of treatment (neoadjuvant or adjuvant).

Fig. 2.

A: Types of grade ≥3 immune-related adverse events (irAEs). B: Distribution of ≥G3 irAEs by system.

3.3. Overall management of irAEs

Among the patients who presented with irAEs and had data on management available (N = 103), 58 patients (56 %) needed corticosteroids, and 2 patients required additional immunosuppressive therapy with infliximab. Among patients with grade 3 or 4 AE, 68.2 % needed corticosteroids (N = 30). There were eight grade 4 events that required corticosteroids: 3 cases of pneumonitis, adrenal insufficiency, colitis, pancytopenia, and a cutaneous event. Additionally, twenty-four grade 3 events required steroids, the most common being hepatitis, with 10 cases, followed by 5 cases of cutaneous toxicity, and 3 cases of colitis. Among the grade 2 events managed with corticosteroids, the most common was pneumonitis (6 patients). The two events that led to use of additional immunosuppressive therapy with infliximab were pneumonitis and red cell aplasia. At last follow-up, 69.9 % of irAEs had resolved. On the other hand, 31 patients had persistent irAEs, mostly endocrine: thyroiditis (n = 17), adrenal insufficiency (n = 4), hypophysitis (n = 2), and diabetes mellitus (n = 1). Pembrolizumab rechallenge was possible in 53.4 % of cases. Patients who underwent rechallenge had a slightly lower median age compared to the overall cohort, all patients had an ECOG performance status of 0, and the majority experienced grade 1 or 2 toxicity (Supplemental Table 2). Among the patients who were rechallenged, 35 had already completed adjuvant therapy by the time of the study analysis, and only 1 of them required early discontinuation of pembrolizumab. Permanent discontinuation of pembrolizumab was necessary for 59 patients (51.8 % of those who had an irAE and 16 % of all those exposed to pembrolizumab). Among patients who permanently discontinued immunotherapy, 59.3 % experienced grade 3 or 4 events. The main irAEs leading to definitive discontinuation were dermatologic (23.5 %), gastrointestinal (21.6 %), and pulmonary (21.6 %).

3.4. Incidence and management of irAEs in the neoadjuvant phase

During the neoadjuvant phase, 83 patients experienced any grade irAEs (22.6 % of the whole cohort), of whom 48 (13 %) happened during the carboplatin and paclitaxel phase, 39 (10.6 %) during the doxorubicin and cyclophosphamide phase, and 4 patients had events on both phases. Grade ≥3 events occurred in 31 patients (8.4 %), and the most common events were hepatitis (13 patients – 3.5 %), pneumonitis (6–1.6 %), cutaneous (6–1.6 %), and endocrine (3 adrenal insufficiencies, 1 hypophysitis, and 1 thyroiditis). Management of irAEs in the neoadjuvant phase was based on corticosteroid therapy in 45 patients (12.2 %), 1 patient (0.3 %) required additional immunosuppressive therapy, and 28 patients (7.6 %) did not require specific treatment. Immunomediated toxicity was resolved in 57 patients (15.5 %). Immunotherapy rechallenge was possible in 47 patients out of 83 patients (56.7 % of cases). Permanent discontinuation of pembrolizumab during the neoadjuvant phase occurred in 41 patients (11.1 %). The main irAEs responsible for permanent discontinuation during this phase were: gastrointestinal (n = 11), pneumonitis (n = 4), and cutaneous (n = 4).

3.5. Incidence and management of irAEs in the adjuvant phase

In the adjuvant phase, 32 patients (11.3 % of the 284 patients who started on adjuvant pembrolizumab and 18.1 % of those who completed the adjuvant phase) had any grade irAEs related to pembrolizumab, of whom 9 patients had events grade ≥3. The most common grade ≥3 events were endocrine (2 thyroiditis, 1 hypophysitis, 1 adrenal insufficiency and 1 diabetes mellitus), along with 1 case of pneumonitis, 1 colitis, 1 arthritis, and 1 cutaneous toxicity. Thirteen patients required corticosteroids, 15 were observed without specific treatment, and 1 case required immunosuppressant therapy with infliximab. Immunomediated toxicity was resolved in 15 out of 32 patients, 12 cases were re-challenged with immunotherapy, and 17 patients required permanent discontinuation of pembrolizumab. The frequency of irAEs by phase of treatment (neoadjuvant or adjuvant) is described on Fig. 1B.

3.6. Time to onset and duration of irAEs

The median time to onset of irAEs is illustrated on Fig. 3. The events with the earliest median time to onset were gastrointestinal (11.4 weeks), cutaneous (11.7 weeks) and thyroid-related (13.4 weeks). Late-onset irAEs included pneumonitis and musculoskeletal events, with a median onset of 25 and 35 weeks, respectively. The median duration of irAEs was 4.2 weeks (range 0.3–59.7). The irAEs with the longest duration were endocrine and musculoskeletal events, both with a median duration of 12 weeks. The events with the quickest resolution were cutaneous and pulmonary, with median duration of 3 and 2.8 weeks, respectively. The 24-week and 48-week irAE-free survival was 80.7 % and 69.9 % (Fig. 4).

Fig. 3.

Median time to onset of the main immune-related adverse events associated with pembrolizumab.

Fig. 4.

Immune-related adverse events-free survival curve.

3.7. Relationship between irAEs and pathological complete response

The pathological complete response (pCR) rate among patients with irAEs was 66.1 %, compared to 60.8 % in patients without irAEs. No statistically significant association was observed between incidence of irAEs and pCR status (p = 0.402) (Fig. 5). Considering only irAEs during the neoadjuvant phase, 59.2 % of patients with neoadjuvant irAE had pCR, compared to 63.3 % in patients without neoadjuvant irAEs (p = 0.582). Furthermore, we did not find a statistically significant difference between pCR among patients with and without thyroid-related immunomediated events (66.7 vs. 62 %, respectively).

Fig. 5.

Relationship between immune-related adverse events (irAEs) and pathological complete response (pCR) rate.

4. Discussion

The real-world data reported in this manuscript from the Neo-Real study provide insights into the patterns of irAEs associated with the use of pembrolizumab in non-metastatic TNBC. Our findings are consistent with those described in the pivotal KEYNOTE-522 trial [3,4,6]. In this trial, the incidence of irAEs in the arm receiving immunotherapy was 35 %, with 13 % of grade ≥3 events, comparable to the incidence in our cohort: 31 % of irAEs (114 patients out of 368), with 13.6 % experiencing grade 3 or higher events. The proportion of irAEs by affected system was also similar, with endocrine, cutaneous, and gastrointestinal being the most common in both our real-world cohort and in KEYNOTE-522 trial. Although our population did not have any deaths from irAEs, there were 2 deaths in KEYNOTE-522 due to irAEs (pneumonitis and autoimmune encephalitis). Randomized controlled trials (RCTs) alone may not provide a comprehensive description of adverse events that often emerge in routine clinical practice, emphasizing the importance of generating real-world data, which includes a broader spectrum of patients and better reflects daily clinical practice [9].

This study is one of the largest real-world cohort of patients treated according to the KEYNOTE-522 regimen, including 368 patients from 10 different institutions. A recently published study reported the efficacy and toxicity of pembrolizumab in 100 early TNBC patients treated at a single center in France, with a 61 % incidence of irAEs (30 % grade ≥3 events), which is higher than the incidence found in our cohort [13]. In the French study, a significant portion of irAEs also occurred in the adjuvant phase (34 %). An abstract presented at the 2024 ASCO Annual meeting reported real-world data from two North American institutions, with a total of 233 patients receiving the KEYNOTE-522 regimen who were not enrolled in clinical trials. The overall incidence of irAEs was 34 %, and the most common irAEs were endocrine (52 %) and gastrointestinal (23 %) [14]. Another abstract published at the same conference described the treatment experience of 106 patients at a single North American institution, with a lower incidence of pembrolizumab-related adverse events, around 17 % [15]. Two other retrospective analyses (presented at the San Antonio Breast Cancer Symposium 2023) with 51 and 79 patients treated according to KEYNOTE-522 showed irAE incidence of 29.4 % and 35.7 %, respectively, similar to the frequency found in our study [16,17].

Previous studies whose population were predominantly patients with lung carcinoma and melanoma have found an association between high body mass index (BMI) [18], smoking status and older age [19] with higher rates of irAEs. In our report, we did not find any baseline characteristic (age, histologic type, histologic grade, tumor stage, body mass index, type of AC regimen, menopausal status, and use of granulocyte colony-stimulating factor) associated with the incidence of irAEs in the logistic regression. Two other cohorts evaluating real-world pembrolizumab data in early TNBC with 233 and 35 patients also found no relationship between clinical factors and the development of irAEs [14,20]. We postulate that the risk factors for the development of irAEs, as well as the relationship with efficacy outcomes, may vary depending on the primary tumor site for which immunotherapy is directed.

Endocrine irAEs were the most common in our analysis (12.8 %), with most events being thyroiditis (9.5 %), and adrenal insufficiency (1.9 %). They represented approximately 20 % of all grade ≥3 irAEs, had the longest median duration, and accounted for most persistent irAEs (24 out of 31). These findings reinforce that patients undergoing immunotherapy should be closely monitored for proper identification and management of these events, as such persistent endocrine toxicities may lead to lifelong sequelae [21,22]. In our analysis, the incidence of adrenal insufficiency (primary + hypophysitis) was less than 5 % of the total evaluated population. Differently, an exploratory analysis of a phase II multicenter study (I-SPY2) with 442 patients receiving different neoadjuvant immunotherapy regimens (including drug combinations) for early TNBC showed a higher incidence (11.8 %) of adrenal insufficiency (primary + hypophysitis) [23]. Another real-world cohort with pembrolizumab in early TNBC reported 8 % of hypophysitis and 6 % of primary adrenal insufficiency [13]. It is important to evaluate why these differences emerged among this distinct patient populations.

Notably, 18.1 % of patients who completed the adjuvant phase of pembrolizumab in our cohort experienced any grade irAEs during this phase of treatment. Given the uncertainty surrounding the benefit of adjuvant pembrolizumab, particularly in patients who achieve pCR, the risks of irAEs, many of which are irreversible, must be carefully weighed against potential benefits. This issue is expected to be further explored by the ongoing OptimICE-pCR trial (NCT05812807), which compares pembrolizumab to observation in patients with pCR following preoperative pembrolizumab combined with chemotherapy [24].

Management of irAEs varies depending on the toxicity grade, but their prompt recognition is crucial for a rapid resolution. If a particular irAE does not resolve within 48–72 h of corticosteroid use, other immunosuppressive agents such as infliximab and mycophenolate mofetil are recommended [25,26]. Among patients managed for irAEs in our study, 56 % received corticosteroid therapy, the same rate as the cohort of Jayan et al. [14], although in that sample 15 patients required additional immunosuppressive agents, compared with only 2 in our population. The rate of discontinuation due to adverse events in KEYNOTE-522 in the pembrolizumab–chemotherapy arm was 27.7 %, but there was no specification of what proportion was due to immune-mediated events [4]. Our study found a discontinuation rate of 16 % due to irAEs, higher in comparison with the 3.6 % reported in an aggregated analysis of about 9000 patients who received pembrolizumab monotherapy across 31 clinical trials, in 19 different cancer types (excluding breast cancer) [8].

There is mixed evidence about a possible association between the incidence of immune-related adverse events and efficacy outcomes, such as response rate and survival. A positive association is more robust among patients with lung cancer and melanoma treated with immunotherapy, while data are still scarce for breast cancer [27]. A retrospective analysis of 35 patients who received neoadjuvant pembrolizumab for TNBC found a statistically significant correlation between the occurrence of irAE and pathological complete response, with 72.2 % pCR in patients with irAEs compared to 30.8 % in patients without irAE (p = 0.03) [20]. However, we did not find any association in our cohort, where the pCR rate among patients with irAEs was 66.1 %, compared to 60.8 % in patients without irAEs (p = 0.402).

The limitations of this study reflect the challenges in collecting real-world data from different institutions, with variations in health records, missing data and the biases inherent to retrospective data collection. Moreover, some clinical data, such as information on personal history of autoimmune disease or smoking status, were not reported for the analysis. Nonetheless, our study provides valuable real-world information from a large multicenter population that reinforces the importance of adequate monitoring and management of irAEs in a population of mostly young patients treated with curative intent.

5. Conclusion

In this real-world data analysis, we observed a similar incidence of irAEs as reported in the KEYNOTE-522 trial. Despite most patients experienced resolution of their irAEs, a significant proportion required permanent discontinuation of pembrolizumab, and some will experience lasting dysfunctions, particularly endocrine, demanding lifelong medical interventions. Notably, the rate of irAEs in the adjuvant phase is not neglectable, especially considering that many of them will be permanent. Thus, data on adjuvant pembrolizumab's effectiveness are needed to weigh its necessity against its risks. Careful monitoring and management of these events are essential to ensure the safety of patients undergoing this regimen.

CRediT authorship contribution statement

Matheus de Oliveira Andrade: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Isabella Gonçalves Gutierres: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation. Monique Celeste Tavares: Writing – review & editing, Data curation. Isadora Martins de Sousa: Writing – review & editing, Data curation. Flávia Cavalcanti Balint: Writing – review & editing, Data curation. Ana Carolina Marin Comini: Writing – review & editing, Data curation. Mariana Carvalho Gouveia: Writing – review & editing, Data curation. José Bines: Writing – review & editing, Data curation. Fernanda Madasi: Writing – review & editing, Data curation. Rafael Dal Ponte Ferreira: Writing – review & editing, Data curation. Daniela Dornelles Rosa: Writing – review & editing, Data curation. Candice Lima Santos: Writing – review & editing, Data curation. Daniele Assad-Suzuki: Writing – review & editing, Data curation. Zenaide Silva de Souza: Writing – review & editing, Data curation. Júlio Antônio Pereira de Araújo: Writing – review & editing, Data curation. Débora de Melo Gagliato: Writing – review & editing, Data curation. Carlos Henrique dos Anjos: Writing – review & editing, Data curation. Bruna M. Zucchetti: Writing – review & editing, Data curation. Anezka Ferrari: Writing – review & editing, Data curation. Mayana Lopes de Brito: Writing – review & editing, Data curation. Renata Cangussu: Writing – review & editing, Data curation. Maria Marcela Fernandes Monteiro: Writing – review & editing, Data curation. Paulo M. Hoff: Writing – review & editing, Supervision. Maria Del Pilar Estevez-Diz: Writing – review & editing, Supervision. Laura Testa: Writing – review & editing, Supervision. Romualdo Barroso-Sousa: Writing – review & editing, Writing – original draft, Supervision, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Renata Colombo Bonadio: Writing – review & editing, Writing – original draft, Supervision, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.

Funding source

None.

Declaration of interest statement

The Authors declare no Competing Non-Financial Interests but the following Competing Financial Interests: RCB: Speaker fees and/or honoraria for consulting or advisory functions: Daiichi-Sankyo, Nestle Health Science, Addium, Gilead, MSD, BMS, AstraZeneca, Ache, Pfizer. Financial support for educational programs and symposia: AstraZeneca, Daiichi-Sankyo, MSD, Lilly. Institutional Research grant: Novartis, AstraZeneca. JB: Speaker fees and/or honoraria for consulting or advisory functions: AstraZeneca, Daiichi-Sankyo, Lilly, Gilead, Pfizer, Novartis, MSD, Roche, Knight Pharmaceuticals. Financial support for educational programs and symposia: Roche, Daiichi-Sankyo. DDR: Speaker fees and/or honoraria for consulting or advisory functions: AstraZeneca, Daiichi-Sankyo, Lilly, Libbs, Pfizer, Novartis, Roche, GSK, Sanofi, Amgen, Zodiac Pharma. Financial support for educational programs and symposia: Roche. DAS: Speaker fees and/or honoraria for consulting or advisory functions: Daiichi-Sankyo. Financial support for educational programs and symposia: AstraZeneca. JAPA: Speaker fees and/or honoraria for consulting or advisory functions: Novartis, AstraZeneca, MSD, Lilly. DMG: Speaker fees and/or honoraria for consulting or advisory functions: Daiichi-Sankyo, Teva, Roche, AstraZeneca, Pfizer, Lilly, Novartis. Financial support for educational programs and symposia: AstraZeneca, Libbs, Roche. Research grant: Novartis. BMZ: AstraZeneca, Daiichi-Sankyo, Eli Lilly, Gilead, Pfizer, Novartis, MSD, Roche, Addium. AF: Speaker fees and/or honoraria for consulting or advisory functions: Daiichi-Sankyo, Novartis, Gilead, MSD, BMS, AstraZeneca, Pfizer. Financial support for educational programs and symposia: AstraZeneca, Daiichi-Sankyo, MSD, Novartis. CHA: Speaker fees and/or honoraria for consulting or advisory functions: Daiichi-Sankyo, Gilead AstraZeneca, Novartis, MSD. Financial support for educational programs and symposia: AstraZeneca, Daiichi-Sankyo, MSD, Lilly, Rcohe, Novartis, Gilead, Medscape. RC: Speaker fees and/or honoraria for consulting or advisory functions: AstraZeneca, Daichii-Sankyo, Eli Lilly, Gilead, MSD, and GSK. MMFM: Speaker fees and/or honoraria for consulting or advisory functions: AstraZeneca, Daiichi-Sankyo, Eli Lilly, Gilead, Adium, Novartis, MSD, and Roche. Financial support for educational programs and symposia: AstraZeneca, Daiichi-Sankyo, Gilead, Eli Lilly, Roche, MSD and Novartis. PMH: Speaker fees and/or honoraria for consulting or advisory functions: Daiichi-Sankyo. MPED: Speaker fees and/or honoraria for consulting or advisory functions: AstraZeneca. LT: Speaker fees and/or honoraria for consulting or advisory functions: Daiichi-Sankyo, MSD, AstraZeneca, Pfizer, Lilly, Novartis, Roche, Pfizer. Financial support for educational programs and symposia: AstraZeneca, Roche, Gilead, MSD. Institutional Research grant: Novartis. RBS: Speaker fees and/or honoraria for consulting or advisory functions: AstraZeneca, Daiichi-Sankyo, Eli Lilly, Gilead, Libbs, Pfizer, Novartis, MSD, and Roche. Financial support for educational programs and symposia: AstraZeneca, Daiichi-Sankyo, Gilead, Eli Lilly, and MSD. Institutional Research grant: AstraZeneca, Daiichi-Sankyo. MCG: Speaker fees and/or honoraria for consulting or advisory functions: GSK, Novartis, Knight Therapeutics, MSD. MOA, IGG, IMS, FCB, ACMC, MCT, FM, RPF, CLS, and ZSS declare no conflict of interest.

Appendix A. Supplementary data

The following is the supplementary data to this article: