Patient’s general features
The patients’ clinical characteristics are described in Table 2. The mean age was 63 years [23–87], and most of the patients were male (n=45, 61%). NSCLC and melanoma were the most common tumors (n=39, 53% and n=25, 34%, respectively), while 10 patients (13%) had other cancer types. Notably, three patients exhibited clear cell renal cell carcinomas, two urothelial carcinomas, two colorectal cancers, one breast cancer, one Merkel carcinoma, and one head and neck carcinoma. Patients received a median of two systemic treatments before the first ICPi, which was discontinued in most of cases because of either disease progression (n=56, 76%), toxicity (n=9, 12%), or per protocol for patients included in clinical trials (n=9, 12%). Approximately 72% of patients received at least 1 systemic agent with or without a local treatment, such as radiotherapy or surgery, between the two ICPis courses (Table S1).
Table 2: Patients’ general features
Clinical feature
|
N = 74
|
Mean age [range]
|
63 [23 – 87]
|
Gender N (%) (MD = 5)
|
|
Male
|
45 (61)
|
Female
|
26 (35)
|
Mean number of previous treatments (MD = 9)
|
2
|
Histology N(%)
|
|
Non-Small-Cell Lung cancer
|
39 (53)
|
Melanoma
|
25 (34)
|
Other histology
|
10 (13)
|
Clear Cell Renal Cell Carcinoma
|
3
|
Urothelial Carcinoma
|
2
|
Colon-rectal cancer
|
2
|
Breast cancer
|
1
|
Merkel carcinoma
|
1
|
Head and neck carcinoma
|
1
|
Reason of first ICPi discontinuation N(%)
|
|
Disease progression
|
56 (76)
|
Toxicity
|
9 (12)
|
Per protocol
|
9 (12)
|
Treatments between the two ICPis N(%) (MD = 1)
|
|
None
|
21 (28)
|
Systemic therapy
|
37 (50)
|
Systemic therapy + local treatmenta
|
16 (22)
|
MD= Missing data; a: radiotherapy or surgical resection
Response under Re-challenge
Best response achieved upon either the first or second ICPi course was available for 73 patients (Fig. 2). In the overall population, a higher ORR (46% versus 24%, p = 4.10-3) and higher DCR (73% versus 52%, p = 7.10-3) were obtained upon the first course of anti-PD-1 or anti-PD-L1 as compared to re-challenge. Focusing on NSCLC, a higher ORR (41% versus 15%, p = 1.10-2) and higher DCR (64% versus 38%, p = 1.10-2) were confirmed upon the first ICPi. In melanoma patients, neither the ORR (44% versus 32%, p = 3.10-1) nor the DCR (76% versus 64%, p = 3.10-1) were statistically different between the two ICPis courses. However, no association was found between the best response achieved during the first ICPi exposition and that obtained at re-challenge in the whole population (p = 3.10-1).
Progression-free survival under re-challenge
Paired PFS1 and PFSR were evaluable in 61 patients. As expected, the PFS1 was longer than the PFSR (6.6 [95% CI: 4.9–8.4] versus 2.8 months [95% CI: 2.0–4.0], HR 0.57 [95% CI: 0.00–0.87], p = 2.10-3) (Fig. 3). The median PFS1 was 5.6 months [95% CI: 2.8–8.9] in NSCLC patients, 6 months [95% CI 3.9–8.4] in melanoma patients, and 15.8 months [95% CI: 8–22.9] in other cancer types. Interestingly, the median PFSR was quite similar among the NSCLC and melanoma patients at 2.1 months [95% CI: 1.6–3.3] and 2.7 months [95% CI: 1.8–7.0], respectively, while it reached 6 months [95% CI: 4.1 – 12.9] in the case of other histology. Note that a longer PFS1 can predict a longer PFSR, these values being positively correlated (Pearson correlation coefficient 0.347, p = 6.10-3) (Fig. 4). Patients with a PFS1 ≥12 months were shown to display a longer PFSR compared to those presenting a PFS1 <12 months (5.1 [95% CI: 2.0–11.1] versus 2.4 months [95% CI: 1.7–3.4], p = 4.10-2) (Fig. 4). Shorter cut-offs were also tested (3, 6, and 9 months), whereas no statistically significant differences were obtained. As expected, a longer PFSR was achieved in patients who discontinued the first anti-PD-1/anti-PD-L1 agent due to toxicity or per protocol as compared to those who experienced disease progression (8.8 [95% CI 5.1–15] versus 2.1 months [95% CI: 1.8–3.1], p = 2.10-3), as well as in patients who did not receive another treatment between the two ICPis courses compared to those who did (6.6 months [95% CI: 2.3 – 14.8] versus 2.1 months [95% CI: 1.6 – 3.3], p = 1.10-3) (Fig. 4). Similar results were obtained, without considering the eight patients receiving ipilimumab (anti-CTLA-4) between the two ICPis (data not shown).
Re-challenge safety profile
Concerning toxicities, a higher percentage of all grade adverse events occurred during the first anti-PD-1/anti-PD-L1 course, as compared to re-challenge (n = 58, 78% versus n = 43, 63%, p = 8.10-3) (Table 3). Rash and diarrhea were the most commonly reported adverse events.
Table 3: All grade toxicitiesa during first and second course of anti-PD1 or anti-PDL1
Adverse event n(%)
|
First ICPi
|
Re-challenge
|
Total
|
58(78)
|
43(63)
|
Fever
|
7 (10)
|
4 (5)
|
Appetite loss
|
3 (4)
|
4 (5)
|
Hepatitis
|
4 (5)
|
5 (7)
|
Pneumonitis
|
5 (7)
|
1 (1)
|
Mucositis
|
3 (4)
|
1 (1)
|
Fatigue
|
6 (8)
|
4 (5)
|
Rash
|
8 (11)
|
7 (10)
|
Diarrhea
|
7 (10)
|
8 (11)
|
Nausea/vomiting
|
1 (1)
|
3 (4)
|
a: Toxicities occurring in ≥ 5% of the overall population either during the first or the second course of anti-PD1 or anti-PDL1 have been detailed