Immune-related adverse events of anti-PD-1 immune checkpoint inhibitors: a single center experience

Objectives Immune checkpoint inhibitors (ICIs) stimulate antitumor immune responses and, in parallel, they might trigger autoimmune and other immunopathological mechanisms eventually leading to immune-related adverse events (irAE). In our study, we assessed patients with malignancies who underwent anti-PD-1 treatment at the University of Debrecen, Clinical Center. Patients and methods Between June 2017 and May 2021, 207 patients started ICI treatment at our university. A total of 157 patients received nivolumab and 50 were treated with pembrolizumab. We looked for factors associated with the development of irAEs. In addition to correlation studies, we performed binary logistic regression analysis to determine, which factors were associated with irAEs. We also performed Forward Likelihood Ratio (LR) analysis to determine independent prognostic factors. Results At the time of data analysis, the mean duration of treatment was 2.03 ± 0.69 years. ROC analysis determined that 9 or more treatment cycles were associated with a significantly higher risk of irAEs. A total of 125 patients received ≥9 treatment cycles. Three times more patients were treated with nivolumab than pembrolizumab. Of the 207 patients, 66 (32%) developed irAEs. Among the 66 patients who developed irAEs, 36 patients (55%) developed one, 23 (35%) developed two, while 7 (10%) developed three irAEs in the same patient. The most common irAEs were thyroid (33 cases), dermatological (25 cases), pneumonia (14 cases) and gastrointestinal complications (13 cases). Patients who developed irAEs received significantly more treatment cycles (21.8 ± 18.7 versus 15.8 ± 17.4; p=0.002) and were younger at the start of treatment (60.7 ± 10.8 versus 63.4 ± 10.1 years; p=0.042) compared to patients without irAEs. Pembrolizumab-treated patients developed more but less severe irAEs compared to those receiving nivolumab. Conclusion ICI treatment is very effective, however, irAEs may develop. These irAEs might be related to the number of treatment cycles and the type of treated malignancy.

Based on the mode of action of ICIs described above, the stimulation of antitumor immune responses may, in parallel, result in the enhancement of autoimmune and other immunological pathways and thus the possible development of immune-related adverse events (irAE) of these drugs (3,5,(7)(8)(9)(10)(11)(12).Such irAEs occur in up to 40% of cases receiving ICI monotherapy (5, 10).While anti-CTLA4 + anti-PD-1 combination therapy result in higher response rates and longer progression-free survival than either agent alone, combination therapy has been associated with more frequent irAEs (up to 95%) (4,10,13).Usually irAEs with anti-PD-1 antibodies are less frequent than those with anti-CTLA4 (12).
There have been several recommendations for the monitoring and management of ICI irAEs.IrAEs associated with anti-PD-1 therapeutic agents are generally reversible and well tolerated (5, 16,17).It is also possible that patients who previously received ICI therapy would develop late-onset irAEs (5, 10, 17).The management of such irAEs highly depends on the grade (G) of severity.In mild cases (Grade 1), except for cardiac and neurologic side effects, only symptomatic treatment (NSAIDs, corticosteroids) is required, and ICI treatment could be continued.In cases of moderate (Grade 2) irAEs, oral corticosteroid treatment is necessary with close monitoring of the symptoms.Grade 3 and 4 irAEs might occur in 20-25% of patients undergoing anti-PD-1 treatment and respiratory, and gastrointestinal irAEs are the most frequent among serious events.In cases of severe (Grade 3) irAEs, ICI therapy needs to be temporarily interrupted along with administering parenteral corticosteroids.ICI therapy may be restarted when the symptoms resolved to Grade 1. Finally, ICI therapy should be terminated permanently in more severe and life-threatening cases (Grade 4), and high-dose parenteral corticosteroids or even synthetic or biologic immunosuppressive drugs can be initiated.The management of these irAEs also require a multidisciplinary approach and consultations with other medical specialties, as well as health professionals and advocacy experts (5, 14,[16][17][18][19][20][21][22].
The present study assessed irAEs in patients with malignant solid tumors with anti-PD-1 therapy, either nivolumab or pembrolizumab treatment between 2017 and 2021 at the University of Debrecen.We evaluated the frequency of irAEs, compared these irAEs in nivolumab-versus pembrolizumabtreated patients, and investigated the determinants of irAE development in these patients.To the best of our knowledge, this is the first Hungarian cohort where ICI irAE data were collected and systematically analyzed.

Patients
Between June 2017 and May 2021, ICI treatment was initiated for 207 patients at the Departments of Oncology and Pulmonology, University of Debrecen.Patient characteristics are included in Table 1.Among the 207 patients, there were 138 males and 69 females.Their mean age was 64.6 ± 8.2 years and their age at the initiation of ICI therapy was 62.6 ± 9.8 years (Table 1).Eventually 157 patients received nivolumab and 50 received pembrolizumab (Table 1).At the time of ICI treatment, patients did not receive any additional chemotherapy or radiotherapy.All patients underwent regular follow-ups until the date of data cut, December 31, 2021.

Data collection and statistical analysis
During data collection, we reviewed the charts of all patients and logged all necessary data into an Excel sheet.Statistical analysis was performed using SPSS version 26 (IBM, Armonk, NY, USA) software.Data are expressed as the mean ± SD for continuous variables and percentages for categorical variables.The distribution of continuous variables was evaluated by the Kolmogorov-Smirnov test.As the distribution of data was not normal, non-parametric tests were used.Continuous variables were compared between groups by the Mann-Whitney test, while nominal variables were compared using the c 2 or Fisher's exact test, as appropriate.Correlations of any two continuous variables were determined by the Spearman's test.Binary logistic regression analysis was performed to assess prognostic factors for irAEs.Moreover, we analyzed Forward Likelihood Ratio (LR) to determine independent prognostic factors.Receiver Operating Characteristic (ROC) curves show the sensitivity and specificity for every possible cut-offs for a test.P values < 0.05 were considered statistically significant.
Patients received a mean of 13.9 ± 12.2 cycles of therapy.Altogether 28 patients received 9 or more treatment cycles (Table 1).Among patients receiving pembrolizumab, the most frequent tumors were lung (n=32) and urinary bladder tumors (n=11) (Figure 1).
Finally, we analyzed and compared the nivolumab and pembrolizumab groups.There were three times more patients treated with nivolumab than with pembrolizumab.There were also statistically significant differences in the line of treatment as 88% of nivolumab-treated patients received this ICI in 2 nd line, while pembrolizumab was used as 1 st line treatment in 50% and 2 nd line treatment in 42% of the cases (p<0.01).There were no signifi can t differe nc e s bet wee n the nivolum ab-and pembrolizumab-treated patients with respect to genders, age, age at treatment initiation, treatment duration, number of cycles, the number of patients receiving ≥9 cycles, whether anti-PD-1 treatment was in the past or ongoing, the reasons for discontinuation and PFS (Table 1).Regarding the types of malignancy, 75% of lung and all 34 renal, 14 tonsillo-pharyngeal, 4 esophageal and 4 oral cavity cancer patients received nivolumab.On the other hand, only 25% of lung, as well as all 11 bladder and 3 breast cancer patients, were treated with pembrolizumab (Figure 1).The distribution of indications for nivolumab and pembrolizumab treatment.Numbers show the number of patients with the given type of malignancy.
IrAEs developed after a mean of 10.0 ± 10.4 cycles in anti-PD-1treated patients, which occurred after 12.0 ± 11.8 cycles with nivolumab and 7.0 ± 5.7 cycles with pembrolizumab (p=0.034).If more than one irAEs occurred, the time for the first irAE to appear was calculated (Table 2).
When comparing nivolumab-and pembrolizumab-treated patients, we did not find significant differences in the proportion of patients with irAEs (p=0.078) and in the relative number of different irAEs (p=0.566).When assessing the specific irAEs, nephropathy was significantly more frequent in the pembrolizumab (p=0.010).Otherwise, there were no differences in the various organ-specific irAEs between the two subgroups.Moreover, irAEs developed after significantly more cycles with nivolumab compared to pembrolizumab (p=0.034).Finally, while nivolumab-associated irAEs were almost equally Grade 1 and 2, pembrolizumab treatment resulted in Grade 1 irAEs in 80% of the cases (p=0.027)(Table 2).In the simple Spearman's correlation analysis, the development of irAEs positively and significantly correlated with the length of PFS (R=0.264;p<0.001), the total number of ICI cycles administered (R=0.273;p<0.001) and recent (ongoing) ICI treatment (R=0.183;p=0.008).The number of irAEs also correlated with PFS (R=0.263;p<0.001), the number of ICI cycles (R=0.276;p<0.001) and recent ICI treatment (R=0.193;p=0.005).Finally, the number of ICI cycles administered before the first irAE developed also correlated with PFS (R=0.603;p<0.001) (Table 3).

Factors associated with the development of irAEs
We performed binary regression analysis to determine possible prognostic factors for the development of irAEs.As defined by the ROC analysis (Figure 2), 9 or more treatment cycles as cut-off resulted in an increased risk for irAEs with an Odds ratio (OR) of 3.328 (95%CI: 1.008-1.042;p=0.004), a sensitivity of 72% and a specificity of 54%.The forward LR method also confirmed the same with an OR of 3.578 (95%CI: 1.875-6.831;p<0.001).Nivolumab and pembrolizumab were also compared with respect to the frequency of irAEs.In the binary logistic regression analysis, there was a non-significant tendency showing that pembrolizumab treatment was more frequently associated with irAEs compared to nivolumab (OR: 1.878 [95% CI: 0.980-3.599];p=0.058).However, in the Forward LR analysis, this difference was statistically significant with an OR of 2.169 (95%CI: 1.089-4.321;p=0.028).
Concerning the specific irAEs, in binary comparisons, patients with thyroid irAEs received more treatment cycles than those without thyroid irAEs (23.0 ± 18.8 versus 16.8 ± 17.7; p=0.04).Patients with pneumonitis also received more treatment cycles (23.1 ± 12.0 versus 17.3 ± 18.3; p=0.022) and had a longer duration of treatment compared to those without pneumonitis (2.5 ± 1.2 versus 2.0 ± 0.9 years; p=0.032).We could not identify any associations between dermatological, gastrointestinal or other specific irAEs or other factors studied.

Discussion
ICIs have become a significant breakthrough in the treatment of numerous malignancies (4-8).However, due to their mode of action, irAEs may develop during therapy due to the stimulation of anti-cancer immune responses [reviewed in (3,5,7,(10)(11)(12)21)].As there have been few reports in this field in the Central-Eastern European (CEE) region including Hungary, we aimed to share our experience collected on a relatively large cohort of 207 patients treated with PD-1 inhibitors, either nivolumab or pembrolizumab at the Clinical Center of the University of Debrecen.
In our cohort, only 6 patients needed treatment termination due to irAEs.Eventually one-third of the patients developed at least one  irAE, after a mean 10 treatment cycles.IrAEs can occur early, while late-onset are difficult to predict with available tools and, consequently, are hard to prevent (12).Half of these patients had only one irAE, while one-third of them had two and only 10% had three.In accordance with the literature (3,5,7,(10)(11)(12)21), the most frequent irAEs were thyroid, skin, diseases, pneumonitis and gastrointestinal conditions.We did not observe any myocarditis (23) or neurotoxicity (24) except for one case of polyneuropathy.In general, 60% of the patients developed Grade 1 irAEs.Most irAEs could be well-controlled using internationally accepted oncology and rheumatology protocols (5, 14, 16, 18-21) and national recommendations (5) and did not require treatment discontinuation.Indeed, irAEs with anti-PD-1 are less frequent than those with anti-CTLA4 (12) and in clinical trials the discontinuation rates are 3-8% (12).When comparing the two anti-PD-1 agents, in our study, pembrolizumab was twice more often associated with irAEs compared to nivolumab.On the other hand, regarding severity, nivolumab treatment was associated with relatively less Grade 1 but more Grade 2 irAEs compared to pembrolizumab suggesting that pembrolizumab treatment results in milder irAEs.In most systematic reviews, meta-analyses, and comparative assessments, nivolumab and pembrolizumab had similar safety and tolerability profiles (25)(26)(27)(28).Therefore, the differences found in our study suggesting that pembrolizumab might cause irAEs more often but these irAEs are milder might be due to other conditions.For example, three times more patients were treated with nivolumab compared to pembrolizumab in this cohort.In our study, pembrolizumab was used earlier, more often in 1 st line.Moreover, there were major differences in treatment indications.For example, pembrolizumab was administered to mostly patients with lung cancer.It has not been established, how the underlying malignancy type influences irAE development, severity, and outcomes (3,5,7,10,11,21).Thus, it is difficult to directly compare these two ICIs due to heterogeneity in the treatment environment.
Our results confirmed those from others, suggesting that irAEs might show associations with ICI efficacy (12,15,29,30).Moreover, pneumonitis has been suggested to be predictive of favorable outcomes in patients receiving anti-PD-1 antibodies (12,31).In other studies, risk factors include pre-existing autoimmune diseases, especially those that are active at the time of ICI initiation.In addition, treatment-related factors, such as the type of ICI (anti-PD-1 versus anti-CTLA4), combination of ICIs, as well as intrinsic factors including tumor and genetic heterogeneities, cancer type, tumour microenvironment and the microbiota might also influence the development of autoimmune irAEs (12,32,33).
There have been numerous recommendations for the management and possible prevention of autoimmune irAEs (5, 14,16,[18][19][20][21]34). In our cohort, 60% of irAEs were Grade 1 and most irAEs were easy-to-control and only very few patients required the discontinuation of ICI therapy.As discussed above, many irAEs occur relatively early, in our case, after a mean of 10 treatment cycles.Several preventive strategies and pretreatment assessments of target organ function have been implemented in preventing chemotherapy-related toxicities, which are more predictable than irAEs.With respect to irAEs, no evidence-based algorithms for active surveillance of such events have become available.Most proposed strategies have been based on expert opinion (5,12,16,20).Very few of our patients required cessation of ICI therapy.In most cases, rechallenge after ICI discontinuation is safe and do not lead to repeated irAEs (35).
The strength of our study is that it might be the largest CEE cohort with respect to irAEs associated with ICI therapy.Moreover, we could include a relatively high number of patients from one center and perform multiple analyses to understand the determinants of irAEs.Of course, this study might also have limitations including its single center nature and the solely clinical approach to these issues.

Conclusions
In our cohort of 207 patients treated with nivolumab or pembrolizumab, we achieved a 16-month PFS with both anti-PD-1 agents.One-third of patients developed irAEs, mostly in Grade 1 and did not require treatment discontinuation in all but 6 cases.There were no major differences between the two drugs in general.However, pembrolizumab seemed to be associated with irAEs more frequently, but these irAEs were less severe compared to those of nivolumab, which could be explained by differences in indications, patient numbers, and other factors.Finally, our results also suggest a close relationship between ICI efficacy as determined by PFS and irAEs.Despite the possible limitations of our study, we collected and analyzed data in the CEE region and provided more information on ICI-related irAEs for practicing physicians (34).patient examination, obtaining data GT: laboratory assessments, data provider statistical analysis, data curation All authors work on behalf of the Hungarian OncoRheumatology Network (HORN) initiative.All authors contributed to the article and approved the submitted version.

FIGURE 2 ROC
FIGURE 2ROC analysis of the association of treatment cycles and the development of irAEs.The asterisk indicates the cut-off of 9 cycles.Nine or more treatment cycles as cut-off resulted in an increased risk for irAEs with an Odds ratio (OR) of 3.328 (95%CI: 1.008-1.042;p=0.004), a sensitivity of 72% and a specificity of 54%.

TABLE 1
Clinical characteristics and efficacy results.

TABLE 2
Immune-related adverse events.
*Data are expressed as mean ± SD.Significant differences between the nivolumab versus pembrolizumab groups are in bold italics.Abbreviation: irAE, immune-related adverse event.

TABLE 3
Results of Spearman's correlation analysis: significant correlations.