relapsed/metastatic hnscc: systemic ......relapsed/metastatic hnscc: systemic treatment and...
TRANSCRIPT
RELAPSED/METASTATIC HNSCC: SYSTEMIC TREATMENT AND IMMUNOTHERAPY Prof. Viktor Grünwald
West-German Cancer Center
University Hospital Essen, Germany
CONFLICT OF INTEREST
1. Appointment University Hospital Essen
2. Consultant BMS, MSD, Roche, Ipsen, Novartis, Roche, Merck KGa, Nanobiotix, Janssen, EUSA Pharma
3. Stocks Astra Zeneca, BMS, MSD
4. Patent none
5. Honoraria Astra Zeneca, BMS, MSD, Merck KGa, Eisai, Roche, Ipsen, Novartis, Janssen, Pfizer
6. Financial research support Pfizer (Wyeth), BMS, MSD, Novartis, Astra Zeneca
7. Other financial support MSD, BMS, Merck KGa, BMS, Ipsen
RM-SCCHN - A MIXED POPULATION WITH DIFFERENT OPTION
MILESTONES IN PALLIATIVE TREATMENT OF RM-SCCHN
1985 1992 1994
MTX vs.
CDDP
CDDP/5FU vs.
Carbo/5FU vs.
MTX
CABOvs.
CDDP vs.
CDDP/5FU2006TPF
2008
Platin/5FU vs.
Platin/5FU/CET
Grose WE et al. Cancer Treat Rep 1985;69:577–81. Forastiere AA et al. J Clin Oncol 1992;10:1245–51. Clavel M et al. Ann Oncol 1994;5:521–6. Gibson MK et al. J Clin Oncol 2005;23:3562–7. Baghi M et al. Anticancer Res 2006;26:585–90. Vermorken JB et al. N Engl J Med 2008;359:1116–27. Ferris et al. (2016). NEJM, NEJMoa1602252. Cohen E. W. et al. Lancet 1–12 (2018). doi:10.1016/S0140-6736(18)31999-8.
1977CDDP
2017Nivolumab
2018Pembrolizumab
LONG-TERM SURVIVAL IS RARE WITH CONVENTIONAL CHEMO1st line chemotherapy trials
Vermorken et al. NEJM, 359(11), 1116–1127. http://doi.org/10.1056/NEJMoa0802656
Study period: 2004-2005
Knödler et al. ESMO 2014, Abstract 7328
Median (months)
95% CI
― Arm A DPFC
9.4 8.2 – 11.4
― Arm B PFC
11.6 7.3 – 15.8
HR 1.25 (95% CI, 0.87-1.81)
Log-rank p = 0.2
2010-2014 2016-2019
TPExtreme ASCO `19
2ND LINE IN RM-SCCHN HAS POOR PROGNOSISLimited activity of chemotherapy
Lala, M., Chirovsky, D., Cheng, J. D. & Mayawala, K. Oral Oncol 84, 108–120 (2018).Data sources and software
The PubMed, Cochrane, and Embase databases were searched foreligible trials; trials referenced in the NCCN v1.2016 guidelines werealso explored. Microsoft Office Excel was used to synthesize study re-cords. Where necessary, trial eligibility criteria were compared against
the criteria listed on ClinicalTrials.gov. Meta-analyses of ORR wereconducted in R (version 3.1.3) using the meta for package [9]. Quali-tative graphical analyses of ORR and DOR across identified trials wereperformed using MATLAB (R2012b; MathWorks®, Natick, MA, USA).
A
B
2L = second-line setting. 3L = third-line setting. CI = confidence interval. ORR = overall response rate.
*Langer et al. [24] reported N = 0 responders with paclitaxel treatment. In order to compute 95% CI, a half a patient was added to the analysis.
Fig. 2. Summary of ORRs reported in the 20 tier 1–2b studies included in the systematic review. (A) Forest plot of ORRs and computed 95% CIs. (B) Graphicalrepresentation of the ORRs; the size of the bubble is proportional to the study size (all-patients-as-treated population), and the color of the bubble shows the line oftherapy and/or tier.
M. Lala et al. 2UDO�2QFRORJ\��������������²���
���
Machiels et al. (2015). The Lancet Oncology, 16(5), 583–594. http://doi.org/10.1016/S1470-2045(15)70124-5
Articles
588 www.thelancet.com/oncology Vol 16 May 2015
with independent medical writing assistance supported fi nancially by the funder. All authors were responsible for the report’s content, were involved at all stages of its development, and approved the fi nal version. The corresponding author had responsibility for the fi nal decision to submit for publication.
ResultsWe screened 593 patients for eligibility, and between Jan 10, 2012, and Dec 12, 2013, we enrolled 483 patients
and randomly assigned them to receive afatinib (n=322) or methotrexate (n=161); reasons for patient ineligibility for randomisation are detailed in fi gure 1. Two patients in the afatinib group (one did not meet eligibility criteria and one withdrew consent after randomisation) and one in the methotrexate group (decided not to take the study drug) did not receive treatment. All randomised patients were included in the effi cacy analyses; 480 patients were included in the safety population. Table 1 shows the baseline characteristics of the treatment groups.
Figure 2: Survival outcomes, tumour shrinkage, and subgroup analysis of progression-free survival(A) Kaplan-Meier estimates of progression-free survival. (B) Kaplan-Meier estimates of overall survival. (C) Waterfall plot of maximum percentage tumour shrinkage. Dotted lines show the ≥20% increase and ≥30% decrease cutoffs. (D) Forest plot of progression-free survival according to predefi ned subgroups. PFS=progression-free survival. HR=hazard ratio. OS=overall survival. ECOG=Eastern Cooperative Oncology Group. HNSCC=squamous-cell carcinoma of the head and neck. CR=complete response. PR=partial response. SD=stable disease. PD=progressive disease. *Stratifi ed log-rank test.
Sorted patient index
Max
imum
dec
reas
e fro
m b
asel
ine (
%)
100
80
60
40
20
0
–80
–60
–40
–20
–100Sorted patient index
22·4%34·8%
Factors
TotalBaseline ECOG performance status 0 1Previous use of EGFR-targeted antibodyfor recurrent or metastatic HNSCC No YesSex Men WomenAge (years) <65 ≥65Region Asia Europe North or Latin AmericanSmoking history (pack-years) <10 ≥10Alcohol consumption (units per week) ≤7 >7Primary tumour site Oral cavity Oropharynx Hypopharynx LarynxRecurrence or metastases Recurrent Metastatic Bothp16 Positive NegativeResponse to previous platinum therapy for recurrent or metastatic HNSCC CR, PR, or SD PD
Number of patients
483
131352
196287
41271
355128
43369
60
87381
37491
136153
93101
16764
241
49208
261146
HR (95% CI)
0·80 (0·65–0·98)
0·73 (0·49–1·10)0·77 (0·60–0·98)
0·63 (0·45–0·88)0·91 (0·70–1·19)
0·74 (0·59–0·92)0·95 (0·55–1·64)
0·79 (0·62–1·01)0·68 (0·45–1·03)
0·62 (0·32–1·20)0·82 (0·64–1·04)0·41 (0·21–0·79)
1·05 (0·66–1·70)0·71 (0·56–0·90)
0·79 (0·62–1·00)0·73 (0·46–1·14)
0·69 (0·46–1·04)0·99 (0·68–1·44)0·78 (0·48–1·25)0·59 (0·38–0·92)
0·59 (0·42–0·84)1·18 (0·65–2·14)0·81 (0·60–1·10)
0·95 (0·51–1·75)0·69 (0·50–0·96)
0·82 (0·62–1·09)0·66 (0·45–0·96)
0·25 1 4
1815129630
013926933220002628161
Time (months)271812 21 24159630
0162817225532203716
1453899294876115161
A D
Prog
ress
ion-
free s
urvi
val (
%)
100
80
60
40
20
0
B
Over
all s
urvi
val (
%)
100
80
60
40
20
0
Number at riskAfatinib
Methotrexate
Number at riskAfatinib
Methotrexate
Afatinib MethotrexatePFS events 275 (85%) 135 (84%)Median PFS, months (95% CI) 2·6 (2·0–2·7) 1·7 (1·5–2·4) HR (95% CI) 0·80 (0·65–0·98)p value* 0·030
Afatinib (n=322)Methotrexate (n=161)
Afatinib MethotrexateOS events 237 (74%) 121 (75%)Median OS, months (95% CI) 6·8 (6·1–7·7) 6·0 (5·2–7·8)HR (95% CI) 0·96 (0·77–1·19)p value* 0·70
C
Favours afatinib Favours methotrexate
Afatinib ≥20% increase (n=53) 0 to <20% increase (n=78) >0 to <30% decrease (n=76) ≥30% decrease (n=36)
Methotrexate ≥20% increase (n=34) 0 to <20% increase (n=50) >0 to <30% decrease (n=26) ≥30% decrease (n=10)
MUTATIONAL BURDEN – BASIS FOR CPI ACTIVITY
Lawrence, et al. (2013). Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nature, 499(7457), 214–218. http://doi.org/10.1038/nature12213
SCCHN ARE INFLAMED TUMORS
Mandal, R., Senbabaoglu, Y., Desrichard, A., Havel, J. J., Dalin, M. G., Riaz, N., et al. (2016). JCI Insight, 1(17), e89829. http://doi.org/10.1172/jci.insight.89829
PD-L1-expression is high in SCCHNFigure 1. Prevalence of quantifiable tumor PD-L1 expression levels across tumor types
Krigsfeld et al. AACR 2017: CT143
PD-L1 SCORES HAVE DIFFERENT COMPOSITION
Immune cell Score (IC)
pos. IC or TC/IC or TC*100
percent
PD-L1 SCORES HAVE DIFFERENT COMPOSITION
Immune cell Score (IC)
pos. IC or TC/IC or TC*100
percent
Tumor proportional score (TPS)
pos. TC/TC*100
percent
PD-L1 SCORES HAVE DIFFERENT COMPOSITION
Immune cell Score (IC)
pos. IC or TC/IC or TC*100
percent
Tumor proportional score (TPS)
pos. TC/TC*100
percent
Combined positivity score (CPS)
pos. cells/TC*100
no unit
A NEW PARADIGM IN CANCER TREATMENTCheckpoint inhibitor therapy
Topalian et al. (2016). Nature reviews Cancer 16: 275-287
CLINICAL TUMOR RESPONSE CAN BE SUBSTANTIALClinical case, treated with pembrolizumab
Laura Q. Chow, ECC 2015
NIVOLUMAB AFTER FAILURE OF PLATINRM-SCCHN after RCT or 1st line platin-based therapy
Ferris et al. (2016). NEJM, NEJMoa1602252. http://doi.org/10.1056/NEJMoa1602252
n engl j med nejm.org 6
T h e n e w e ngl a nd j o u r na l o f m e dic i n e
Ove
rall
Surv
ival
(%)
100
80
90
70
60
40
30
10
50
20
00 3 6 9 12 15 18
Months
A Overall Survival
C Treatment Effect on Overall Survival, According to Subgroup
Hazard ratio for death, 0.70(97.73% CI, 0.51–0.96)
P=0.01
No. at RiskNivolumabStandard
therapy
240121
16787
10942
5217
245
71
00
Standardtherapy
Nivolumab
Prog
ress
ion–
free
Sur
viva
l (%
)
100
80
90
70
60
40
30
10
50
20
00 3 6 9 12 15 18
Months
Standardtherapy
Nivolumab
No. ofDeaths
No. ofPatients
NivolumabStandard Therapy
240121
13385
7.5 (5.5–9.1)5.1 (4.0–6.0)
36.0 (28.5–43.4)16.6 (8.6–26.8)
Median OverallSurvival
1–Yr OverallSurvival Rate% (95% CI) mo (95% CI)
B Progression-free Survival
Hazard ratio for disease progressionor death, 0.89 (95% CI, 0.70–1.13)
P=0.32
No. at RiskNivolumabStandard
therapy
240121
7943
329
122
40
10
00
No. ofEvents
No. ofPatients
NivolumabStandard Therapy
240121
190103
2.0 (1.9–2.1)2.3 (1.9–3.1)
Median Progression-freeSurvival (95% CI)
mo
OverallAge
<65 yr≥65 yr and <75 yr≥75 yr
ECOG performance-status score0≥1Not reported
Previous cetuximab useYesNo
Intended standard therapyCetuximabMethotrexateDocetaxel
Site of primary tumorLarynxOral cavityPharynxOther
No. of previous lines of systemic therapy12≥3
Platinum-refractory disease in contextof primary therapyYesNo
Unstratified Hazard Ratio (95% CI)no. of patients
Subgroup
240
1725612
49190
1
14793
3311988
34108926
1068054
52188
121
76396
2397
1
7447
155254
1567363
584518
2695
Standard TherapyNivolumab
0.50 1.00 2.00 4.00 8.00
Standard Therapy BetterNivolumab Better
0.250.125
0.69 (0.53–0.91)
0.64 (0.45–0.89)0.93 (0.56–1.54)
0.60 (0.30–1.23)0.71 (0.53–0.96)
0.81 (0.57–1.15)0.55 (0.35–0.86)
0.47 (0.22–1.01)0.64 (0.43–0.96)0.82 (0.53–1.28)
0.75 (0.36–1.59)0.73 (0.51–1.07)0.71 (0.42–1.19)
—
—
—
0.72 (0.48–1.07)0.64 (0.40–1.00)0.77 (0.38–1.57)
0.63 (0.35–1.12)0.71 (0.52–0.97)
The New England Journal of Medicine Downloaded from nejm.org at MHH-BIBLIOTHEK on October 19, 2016. For personal use only. No other uses without permission.
Copyright © 2016 Massachusetts Medical Society. All rights reserved.
n engl j med nejm.org 6
T h e n e w e ngl a nd j o u r na l o f m e dic i n e
Ove
rall
Surv
ival
(%)
100
80
90
70
60
40
30
10
50
20
00 3 6 9 12 15 18
Months
A Overall Survival
C Treatment Effect on Overall Survival, According to Subgroup
Hazard ratio for death, 0.70(97.73% CI, 0.51–0.96)
P=0.01
No. at RiskNivolumabStandard
therapy
240121
16787
10942
5217
245
71
00
Standardtherapy
Nivolumab
Prog
ress
ion–
free
Sur
viva
l (%
)
100
80
90
70
60
40
30
10
50
20
00 3 6 9 12 15 18
Months
Standardtherapy
Nivolumab
No. ofDeaths
No. ofPatients
NivolumabStandard Therapy
240121
13385
7.5 (5.5–9.1)5.1 (4.0–6.0)
36.0 (28.5–43.4)16.6 (8.6–26.8)
Median OverallSurvival
1–Yr OverallSurvival Rate% (95% CI) mo (95% CI)
B Progression-free Survival
Hazard ratio for disease progressionor death, 0.89 (95% CI, 0.70–1.13)
P=0.32
No. at RiskNivolumabStandard
therapy
240121
7943
329
122
40
10
00
No. ofEvents
No. ofPatients
NivolumabStandard Therapy
240121
190103
2.0 (1.9–2.1)2.3 (1.9–3.1)
Median Progression-freeSurvival (95% CI)
mo
OverallAge
<65 yr≥65 yr and <75 yr≥75 yr
ECOG performance-status score0≥1Not reported
Previous cetuximab useYesNo
Intended standard therapyCetuximabMethotrexateDocetaxel
Site of primary tumorLarynxOral cavityPharynxOther
No. of previous lines of systemic therapy12≥3
Platinum-refractory disease in contextof primary therapyYesNo
Unstratified Hazard Ratio (95% CI)no. of patients
Subgroup
240
1725612
49190
1
14793
3311988
3410892
6
1068054
52188
121
7639
6
2397
1
7447
155254
156736
3
584518
2695
Standard TherapyNivolumab
0.50 1.00 2.00 4.00 8.00
Standard Therapy BetterNivolumab Better
0.250.125
0.69 (0.53–0.91)
0.64 (0.45–0.89)0.93 (0.56–1.54)
0.60 (0.30–1.23)0.71 (0.53–0.96)
0.81 (0.57–1.15)0.55 (0.35–0.86)
0.47 (0.22–1.01)0.64 (0.43–0.96)0.82 (0.53–1.28)
0.75 (0.36–1.59)0.73 (0.51–1.07)0.71 (0.42–1.19)
—
—
—
0.72 (0.48–1.07)0.64 (0.40–1.00)0.77 (0.38–1.57)
0.63 (0.35–1.12)0.71 (0.52–0.97)
The New England Journal of Medicine Downloaded from nejm.org at MHH-BIBLIOTHEK on October 19, 2016. For personal use only. No other uses without permission.
Copyright © 2016 Massachusetts Medical Society. All rights reserved.
2-YEAR FOLLOW-UP WITH SUSTAINED BENEFIT
Ferris, R. L. et al., Oral Oncol. 2018; 81, 45–51
analysis, 8 (3.4%) patients remained on treatment in the nivolumabarm compared with 0 in the IC arm (Supplementary Table 1 and Sup-plementary Fig. 1). Sixteen patients in the nivolumab arm discontinuedtreatment between 1 and 2 years, most commonly due to disease pro-gression (n=8); 2 patients discontinued nivolumab therapy after2 years (1 each due to adverse events unrelated to study drug and pa-tient request). No patient discontinued nivolumab therapy after 2 yearsdue to disease progression or treatment-related toxicity. The median(range) duration of treatment was 1.9 (0 to 36+) months for nivolumaband 1.9 (0 to 13) months for IC. After treatment discontinuation, 5.3%of patients in the nivolumab arm and 10.1% of patients in the IC armreceived subsequent immunotherapy (nivolumab, pembrolizumab,durvalumab, or urelumab) (Supplementary Table 2).
Efficacy in overall study population
Consistent with the primary analysis, nivolumab demonstratedsustained OS benefit compared with IC, with a 32% reduction in risk ofdeath; HR=0.68 (95% CI 0.54–0.86) with long-term follow-up(minimum 24.2 months). Median (95% CI) OS was 7.7 (5.7–8.8)months in the nivolumab arm and 5.1 (4.0–6.2) months in the IC arm(Fig. 1). The Kaplan-Meier–estimated 24-month OS rate with nivo-lumab (16.9% [95% CI 12.4%–22.0%]) was nearly triple that of IC(6.0% [95% CI 2.7%–11.3%]). The estimates of the HR for death amongprespecified demographic and clinical subgroups, including age,Eastern Cooperative Oncology Group (ECOG) performance status,tumor site, and prior lines of systemic therapy, were generally con-sistent with the ITT population, favoring nivolumab (SupplementaryFig. 2). With longer follow-up, PFS (HR=0.87 [95% CI 0.68–1.11])was similar to previous analyses [8,10].
The ORR was unchanged from previous analyses (Table 2) [8,10]. Inthe nivolumab arm, 7 complete responses were observed, including 1patient who had a partial response at the previous analysis, but sincethen converted to a complete response. The median (range) time toresponse remained unchanged in both treatment arms from previous
follow-ups, 2.1 (1.8 to 7.4) months for nivolumab vs 2.0 (1.9 to 4.6)months for IC [8]. The median (range) duration of response with ni-volumab treatment was more than double that with IC (9.7 [2.8 to32.8+] vs 4.0 [1.5+ to 11.3]).
Tumor PD-L1 expression and HPV status
OS benefit with nivolumab vs IC was demonstrated across PD-L1expressors and non-expressors (Fig. 2). With long-term follow-up, ni-volumab continued to provide OS benefit in PD-L1 expressors, with aconsistent 45% reduction in the risk of death compared with IC(HR=0.55 [95% CI 0.39–0.78]) (Fig. 2A). In PD-L1 non-expressors,nivolumab demonstrated a 27% reduction in the risk of death comparedwith IC (HR=0.73 [95% CI 0.49–1.09] (Fig. 2B). For these patients,HR (95% CI) trended lower as follow-up time increased: 0.83(0.54–1.29) and 0.89 (0.54–1.45) at the 1-year (September 2016 datacutoff, Supplementary Fig. 3A) and primary analysis (December 2015data cutoff, Supplementary Fig. 3B) [8], respectively. Kaplan-Meierestimated OS rates with nivolumab were consistent between PD-L1
12219630 15 18 24 27 30 33 36 39
Months
0
10
20
30
40
50
60
70
80
90
100
OS
(%)
240 169 132 98 78
Median OS(95% Cl), mo
Nivolumab (n = 240)IC (n = 121)
7.7 (5.7–8.8) 0.68(0.54–0.86)5.1 (4.0–6.2)
HR(95% Cl)
57 50 42 37 28121 88 51 32 23 14 10 8 7 4
Nivo IC
No. at risk 151
101
40
00
16.9%
Nivo
IC
6.0%
Fig. 1. Overall survival (OS) with a minimum follow-up of 24.2 months (intent-to-treat population). Symbols represent censored observations; dotted lines indicateOS rate time points. CI, confidence interval; HR, hazard ratio; IC, investigator’s choice; Nivo, nivolumab.
Table 2Best overall response.
Nivolumab(n= 240)
IC (n= 121)
Best overall response, n (%)Complete response 7 (2.9) 1 (0.8)Partial response 25 (10.4) 6 (5.0)Stable disease 55 (22.9) 43 (35.5)Progressive disease 99 (41.3) 42 (34.7)Unable to determine 54 (22.5) 29 (24.0)
ORR,% (95% CI) 13.3 (9.3–18.3) 5.8 (2.4–11.6)Time to response, median (range), mo 2.1 (1.8–7.4) 2.0 (1.9–4.6)Duration of response, median (range), mo 9.7 (2.8 to 32.8+) 4.0 (1.5+ to 11.3)
CI, confidence interval; IC, investigator’s choice; ORR, objective response rate.
R.L. Ferris et al. 2UDO�2QFRORJ\�������������²��
��
COMPARABLE EFFECT FOR PEMBROLIZUMABKN040 trial in RM-SCCHN after platin-failure
Cohen, E. W. et al., Lancet. 2018; 1-12. doi:10.1016/S0140-6736(18)31999-8
Articles
6 www.thelancet.com Published online November 30, 2018 http://dx.doi.org/10.1016/S0140-6736(18)31999-8
group and 148 [82%] of 180 in the standard-of-care group)had died (HR 0·93, 95% CI 0·73–1·17; p=0·2675); median overall survival was 6·5 months (95% CI 5·6–8·8) with pembrolizumab and 7·1 months (5·7–8·1) with standard of care (figure 3D). The nominal, two-sided p value for the interaction of treatment effect and PD-L1 tumour proportion score was 0·015 (figure 2B).
In the intention-to-treat population, 36 of 247 patients in the pembrolizumab group and 25 of 248 in the standard-of-care group had a confirmed or unconfirmed response, resulting in a response rate of 14·6% (95% CI 10·4–19·6) and 10·1% (6·6–14·5), respectively (nominal p=0·0610; appendix). Among the 26 patients in the pembrolizumab group and 18 patients in the standard-of-care group who had a confirmed response, median time to response was 4·5 months (IQR 2·3–6·4) with pembrolizumab and 2·2 months (2·1–3·5) with stan-dard of care. The median duration of response was 18·4 months (95% CI 5·8–18·4) with pembrolizumab and 5·0 months (3·6–18·8) with standard of care (appendix). The proportion of patients who had an objective response in the pembrolizumab group was higher in patients whose tumours expressed PD-L1 than in those whose tumours did not, whereas the proportion in the standard-of-care group was similar regardless of PD-L1 expression (appendix). Duration of response was not affected by PD-L1 expression, although the medians fluctuated because of the low number of responses overall (appendix).
With 442 events of death or disease progression assessed according to RECIST version 1.1 in the total population (218 [88%] of 247 in the pembrolizumab group and 224 [90%] of 248 in the standard of care group), no difference in progression-free survival between treatment groups was observed (figure 4). Median progression-free survival was 2·1 months (95% CI 2·1–2·3) with pembrolizumab and 2·3 months (2·1–2·8) with standard
Figure 2: Overall survival in the intention-to-treat populationKaplan-Meier estimates of overall survival according to treatment group in the total population (A) and forest plot of the overall survival findings in subgroups (B). (A) Tick marks represent patients who had data censored at the last time at which they were known to be alive. (B) All subgroups were prespecified except for previous cetuximab treatment, age (prespecified categories were ≤65 years vs >65 years), and region of enrolment (prespecified categories were east Asia vs the rest of the world). Although not a prespecified subgroup analysis, the PD-L1 combined positive score breakdown of less than 1 versus 1 or higher was included for completeness. *Subgroups are based on what the investigator chose before the patient was randomly allocated to treatment with either pembrolizumab or standard of care (investigators were required to select a standard-of-care therapy for all patients before random allocation should they be allocated to that group). The hazard ratios for death for the comparison of pembrolizumab versus standard-of-care therapy in all subgroups were calculated using a Cox proportional hazards model stratified by the randomisation stratification factors. The interaction of each subgroup with treatment was a post-hoc exploratory analysis done using the likelihood ratio test. The two-sided p values are not adjusted for multiplicity and, therefore, nominal only; small p values suggest that the treatment effect varies across subgroups. ECOG=Eastern Cooperative Oncology Group. HR=hazard ratio. PD-L1=programmed death ligand 1.
Events/patients
Pembrolizumab Standard of care
Hazard ratio(95% CI)
pInteraction value
0·50
0·07
0·05
0·61
0·87
0·47
0·32
0·32
0·63
0·38
0·07
0·015
OverallSexMaleFemaleAge (years)<65≥65 to <75≥75RaceWhiteNon-whiteRegion of enrolmentEuropeNorth AmericaRest of the worldECOG performance status01Smoking statusCurrentFormerNeverp16 status in the oropharynxPositiveNegativePrevious lines of therapyCurative intent1≥2Previous cetuximabYesNoChoice of chemotherapyMethotrexateDocetaxelCetuximabCombined positive score<1≥1Tumour proportion score<50%≥50%
181/247
152/20729/40
126/16540/6315/19
149/20718/24
107/14752/7322/27
44/71137/176
22/32107/147
52/68
44/61137/186
21/34103/141
57/72
109/14572/102
51/7094/12336/54
42/50138/196
139/18241/64
207/248
177/20530/43
141/16758/69
8/12
170/20722/26
141/15838/6028/30
51/67155/180
31/36120/146
56/66
45/58162/190
33/40120/141
54/67
113/14094/108
54/6591/11062/73
42/5462/191
148/18056/65
0·80 (0·65–0·98)
0·77 (0·62–0·96)0·94 (0·54–1·63)
0·94 (0·73–1·20)0·57 (0·37–0·87)1·13 (0·42–3·02
0·80 (0·64–1·00)0·93 (0·45–1·92)
0·68 (0·52–0·88)1·27 (0·82–1·97)0·67 (0·36–1·26)
0·87 (0·57–1·32)0·78 (0·62–0·98)
0·71 (0·38–1·31)0·78 (0·60–1·02)0·90 (0·60–1·35)
0·97 (0·63–1·49)0·77 (0·61–0·97)
0·77 (0·42–1·43)0·77 (0·59–1·00)0·96 (0·65–1·43)
0·89 (0·68–1·16)0·78 (0·56–1·07)
0·75 (0·50–1·13)0·86 (0·64–1·17)0·56 (0·36–0·89)
1·28 (0·80–2·10)0·74 (0·58–0·93)
0·93 (0·73–1·17)0·53 (0·35–0·81)
Favourspembrolizumab
Favoursstandard of care
1·00·1 5·0
Number at risk(number censored)
PembrolizumabStandard of care
0 5 10 15 20 25 30
247 (0)248 (0)
160 (0)151 (3)
48 (33)34 (19)
14 (55)10 (35)
2 (64)1 (40)
103 (2)82 (3)
0 (66)0 (41)
Time (months)
0102030405060708090
100
Over
all s
urvi
val (
%)
B
APembrolizumabStandard of care
HR 0·80 (95%CI 0·65–0·98);nominal p=0·0161
Articles
www.thelancet.com Published online November 30, 2018 http://dx.doi.org/10.1016/S0140-6736(18)31999-8 7
of care. Progression-free survival in the population with a PD-L1 combined positive score of 1 or higher was similar to that of the total population, whereas progression-free survival was longer with pembrolizumab in the population with a PD-L1 tumour proportion score of 50% or higher (figure 4); progression-free survival appeared to be shorter with pembrolizumab than with standard of care in the populations with a combined positive score of less than 1 and tumour proportion score of less than 50%. Median progression-free survival was longer in both treatment groups when assessed according to modified RECIST than when assessed with standard RECIST, and the HRs were close to 1·00 in both the total population and population with a PD-L1 combined positive score of 1 or higher (appendix). No difference in time to progression assessed according to RECIST version 1.1 was observed in either the total population or the population with a PD-L1 combined positive score of 1 or higher (appendix).
In the intention-to-treat population, 84 (34%) of 247 patients in the pembrolizumab group and 101 (41%)
of 248 patients in the standard-of-care group received subsequent therapy, including 11 (4%) of 247 patients in the pembrolizumab group and 31 (13%) of 248 patients in the standard-of-care group who received subsequent therapy with an immune checkpoint inhibitor (appendix). In a post-hoc exploratory analysis in the standard-of-care group, the 31 patients who received subsequent immune checkpoint inhibition had longer overall survival than the 70 patients who received other subsequent therapy and the 147 patients who received no subsequent therapy (med ian overall survival of 20·1 months vs 9·7 months vs 4·5 months; appendix). In a post-hoc sensitivity analysis in which patients in both treatment groups were censored at the time of first subsequent immune checkpoint inhibitor, the HR for death was 0·72 (95% CI 0·58–0·88; nominal p=0·0008; appendix). In this analysis, median over-all survival was 8·3 months (95% CI 6·4–9·4) with pembrolizumab and 6·6 months (5·4–7·5) with stan-dard of care.
Figure 3: Overall survival in the intention-to-treat populations according to PD-L1 expression category Kaplan-Meier estimates of overall survival according to treatment group in the population with a combined positive score of 1 or more (A), the population with a combined positive score of less than 1 (B), the population with a tumour proportion score of 50% or more (C), and the population with a tumour proportion score of less than 50% (D). Tick marks represent patients who had data censored at the last time a which they were known to be alive. HR=hazard ratio. PD-L1=programmed death ligand 1.
PD-L1 combined positive score ≥1
PD-L1 tumour proportion score ≥50%
PD-L1 combined positive score <1
PD-L1 tumour proportion score <50%
PembrolizumabStandard of care
HR 0·74 (95% CI 0·58–0·93);nominal p=0·0049
HR 0·53 (95% CI 0·35–0·81);nominal p=0·0014
HR 1·28 (95% CI 0·80–2·07);nominal p=0·8476
HR 0·93 (95% CI 0·73–1·17);nominal p=0·2675
Number at risk(number censored)
PembrolizumabStandard of care
0 5 10 15 20 25 30
196 (0)191 (0)
131 (0)115 (3)
87 (2)63 (3)
43 (26)28 (13)
14 (47)8 (25)
0 (58)0 (29)
2 (56)1 (28)
0102030405060
708090
100
Over
all s
urvi
val (
%)
0 5 10 15 20 25 30
50 (0)54 (0)
28 (0)35 (0)
15 (0)18 (0)
4 (7)6 (6)
0 (8)2 (10)
0 (8)0 (12)
0 (8)0 (12)
A B
Number at risk(number censored)
PembrolizumabStandard of care
0 5 10 15 20 25 30
64 (0)65 (0)
49 (0)38 (2)
35 (1)22 (2)
19 (8)9 (4)
7 (18)2 (9)
0 (23)0 (9)
1 (22)0 (9)
Time (months)
0102030405060
708090
100
Over
all s
urvi
val (
%)
0 5 10 15 20 25 30
182 (0)180 (0)
110 (0)112 (1)
67 (1)59 (1)
28 (25)25 (15)
7 (37)8 (26)
0 (43)0 (32)
1 (42)1 (31)
Time (months)
C D
SUPPORTIVE DATA FOR USE OF DURVALUMAB (PD-L1I)Strict 2nd line in RM-SCCHN
Siu et al., JAMA Oncol. 2018; 1-9. doi:10.1001/jamaoncol.2018.4628
and CheckMate-141 did enroll all comers in terms of PD-L1 ex-pression; therefore, some patients had high PD-L1 expressionaccording to the specific algorithms used for those studies.
The CONDOR study was not powered to compare the com-bination and monotherapy arms, but addition of tremelim-umab did not appear to improve outcomes compared withsingle-agent durvalumab in patients with PD-L1–low/negativedisease. However, both the combination and durvalumabmonotherapy arms showed clinically relevant OS. In Keynote-040, median OS was 8.4 months for pembrolizumab vs 6.9
months for investigators’ choice of therapy, whereas, in pa-tients with a combined positive score of 1 or higher (PD-L1 TCand immune cell expression ≥1), median OS was 8.7 monthswith pembrolizumab vs 7.1 months with investigators’ choiceof therapy. In patients with a tumor proportion score of greaterthan 50%, median OS was 11.6 months with pembrolizumab vs6.6 months with investigators’ choice of therapy.32 In Check-Mate-141, median OS in response to nivolumab treatment was8.2 months for patients with PD-L1–positive expression (≥1%tumor cell membrane staining) vs 6.5 months for patients withPD-L1–negative expression (<1% tumor cell membranestaining).33 Considering all comers in CheckMate-141, in whichthe median OS in response to nivolumab therapy was 7.7 monthsvs 5.1 months for the investigators’ choice of therapy, our studyshows similar median survival for patients treated with dur-valumab + tremelimumab (7.6 months) or durvalumab alone(6.0 months), indicating their potential utility irrespective ofPD-L1 expression.
The rationale for combining anti–PD-L1 and anti–CTLA-4mAbs has precedence in other tumor types, including PD-L1–low/negative tumors. The combination of nivolumab and ipi-limumab has yielded higher ORRs compared with each mono-therapy in melanoma.34 A dose of 20 mg/kg of durvalumab plus1 mg/kg of tremelimumab was identified as the maximum tol-erated dose for this combination in a phase 1b study in pa-tients with non–small cell lung cancer, which demonstratedantitumor activity in patients with PD-L1–low/negative(TC<25%) expression.15 Higher dosages of tremelimumab wereevaluated in this early-phase study but resulted in a higher fre-quency of TRAEs, grade 3 or 4 adverse events, and serious ad-verse events, without any increase in efficacy (ORR).15 Serumconcentration levels of tremelimumab in combination withdurvalumab did not show any marked difference betweendoses and T-cell proliferation, and activation markers for dif-ferent doses of tremelimumab also did not show any markeddifference in this study.15 It is therefore unlikely that the treme-limumab dose used in the combination arm in the present studylimited treatment efficacy. The lack of efficacy of tremelim-umab may be related to its mechanism of action, which as anIgG2 mAb does not cause lysis of regulatory T cells through an-tibody-dependent cell-mediated cytotoxicity, which is ob-served with ipilimumab.35 The clinical relevance of this dif-ference remains uncertain.
All arms exhibited manageable safety profiles in a patientpopulation with few treatment options. The durvalumabmonotherapy safety profile is consistent with previous data,11,12
and the combination arm produced no additional safety con-cerns. Grade 3 or 4 TRAEs occurred in 15% of patients overalland imAEs reported in the study were typical of the PD-1/PD-L1/CTLA-4 class of immunotherapies, which have previouslybeen associated with hypothyroidism, diarrhea, pneumoni-tis, and colitis.11,12,15,26,27,30,31,36,37 One treatment-related deathwas reported in a patient receiving combination therapy.
LimitationsThe most apparent limitation to this study is that the studywas not powered to compare the combination and mono-therapy arms.
Figure 2. Kaplan-Meier Estimates of Progression-Free Survivaland Overall Survival
0
No. at risk
0
1336767
3
392412
6
1612
1
9
1381
12
550
15
100
18
000
1.00.90.80.70.60.50.40.30.20.1
Prob
abili
ty o
f PFS
, %
Time From Randomization, mo
Durvalumab + tremelimumab
TremelimumabDurvalumab
Progression-free survivalA
0
No. at risk
0
1336767
3
894843
6
693328
9
572721
12
371914
15
1793
18
300
21
000
1.00.90.80.70.60.50.40.30.20.1
Prob
abili
ty o
f OS,
%
Time From Randomization, mo
Durvalumab + tremelimumab
TremelimumabDurvalumab
Overall survivalB
Durvalumab + tremelimumabDurvalumabTremelimumab
Study Arm
2.0 (1.9-2.1)1.9 (1.8-2.8)1.9 (1.8-2.0)
Median (95% CI)PFS, mo
1 [Reference]1.13 (0.82-1.56); P = .470.73 (0.53-1.01); P = .05
HR (95% CI)a;P Value
Durvalumab + tremelimumab
DurvalumabTremelimumab
Study Arm
7.6 (4.9-10.6)
6.0 (4.0-1.3)5.5 (3.9-7.0)
Median (95% CI)OS, mo
6.5
6.05.2
Median Follow-up, mo
1 [Reference]
0.99 (0.69-1.43); P = .890.72 (0.51-1.03); P = .06
HR (95% CI)a;P Value
Hatching on plots indicates censored data.a Stratified log rank test.
Durvalumab Monotherapy and Combination Therapy in PD-L1–Low/Negative HNSCC Original Investigation Research
jamaoncology.com (Reprinted) JAMA Oncology Published online November 1, 2018 E7
© 2018 American Medical Association. All rights reserved.
Downloaded From: by a Universität Duisburg Essen User on 11/02/2018
Zandberg, D. P. et al. Eur J Cancer 107, 142–152 (2019).
Fig. 1. KaplaneMeier estimates of PFS (A) and OS (B). Exploratory analysis of KaplaneMeier estimates of OS by HPV status (C). HPV
status was unknown for 13 patients (full analysis set; NZ 112). CI, confidence interval; HPV, human papillomavirus; OS, overall survival;
PFS, progression-free survival.
D.P. Zandberg et al. / European Journal of Cancer 107 (2019) 142e152 147
CONDOR: PD-L1 ≦ 25% HAWK: PD-L1 ≧ 25%
RESPONSE – A KEY INGREDIENT FOR TREATMENT SUCCESSQuality of response is a putative driver of outcome
0 3
NivoIC
6 9 12 18 2415 21 27
100908070605040302010
0
OS
(%)
Months
CR/PR
32NivoNo. at risk
IC32 30 30 23 7 1 0
7 7327 6 4 1
140 0 0 0
Nivolumab (n = 32)
IC (n = 7)
Median OS (95% CI), monthsa NR (NR, NR) 13.6 (8.9, NR)HR (95% CI)b 0.12 (0.03, 0.51)
12-month OS rate (95% CI), % 96.8 (79.2, 99.5) 57.1 (17.2, 83.7)
18-month OS rate (95% CI), % 86.1 (67.0, 94.6) 38.1 (6.1, 71.6)
aKaplan-Meier estimate of median; bUnstratified Cox proportional hazard ratio of nivolumab to IC. NR = not reached
Licitra et al. ESMO 2017: 1055P
CROSS-OVER IN REAL WORLD MAY IMPROVE SURVIVAL
Soulières et al. – AACR 2018
POST-CPI TREATMENT MAY IMPACT OUTCOME
Behandlung mit CPI 232
Geeignet 82 Patienten *
CPI-Therapieform Monotherapie 45 %
Kombi-Therapie 55 %
SCT Taxan-basiert 56 %
Platin-basiert 37 %
Mtx-basiert 7 %
Cetuximab + Taxan/Platin
50 %
Zahl vorangegangener Behandlungen
2 (zwischen 1-6)
*84 % männlich, medianes Alter: 60 Jahre
ORR unter SCT 31 % **
Komplettremission (CR)
4 % (3 Patienten)
Partielle Remission (PR)
27 % (22 Patienten)
Krankheitskontrolle (DCR)
58 %
** 95 Konfidenzintervall: 21 % - 40 %
Saleh et al. ASCO 2018: abstr. 6015
TODAY’S TREATMENT ALGORITHM IN RM-SCCHN
recurrence ≦ 6 mo. after platin
Nivolumab1individual
chemotherapy
Platin-unfit
PF-C2
recurrence > 6 mo. after platin
RM-SCCHN
1.Ferris et al. N Engl J Med 375, NEJMoa1602252–1867 (2016). 2. Vermorken et al. N Engl J Med 359, 1116–1127 (2008).
*TPS ≧ 50%
TODAY’S TREATMENT ALGORITHM IN RM-SCCHN
recurrence ≦ 6 mo. after platin
Nivolumab1individual
chemotherapy
Platin-unfit
PF-C2
recurrence > 6 mo. after platin
RM-SCCHN
1.Ferris et al. N Engl J Med 375, NEJMoa1602252–1867 (2016). 2. Vermorken et al. N Engl J Med 359, 1116–1127 (2008).
individualchemotherapy
Nivolumabor.pembrolizumab*
Nivolumaborpembrolizumab*2nd line
1st line
*TPS ≧ 50%
THE WIND OF CHANGECPI will be part of our 1st line strategy in near futureRM-SCCHN: CPI ist CTX überlegen, alleine oder in Kombination
Overall Survival: P vs E, CPS ≥20 Population
Data cutoff date: Jun 13, 2018.
Events HR (95% CI) PPembro alone 62% 0.61 (0.45-0.83) 0.0007EXTREME 78%
Median (95% CI)14.9 mo (11.6-21.5)10.7 mo (8.8-12.8)
0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 00
1 0
2 0
3 0
4 0
5 0
6 0
7 0
8 0
9 0
1 0 0
M o n th s
OS
, %
N o . a t R is k133 106 85 65 24122 100 64 42 12
4722
00
115
20
12-mo rate56.9%44.9% 24-mo rate
38.3%22.1%
Burtness KN048 ESMO 2018
0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 00
1 0
2 0
3 0
4 0
5 0
6 0
7 0
8 0
9 0
1 0 0
M o n th s
OS
, %
N o . a t R is k281 227 169 122 40278 227 147 100 20
7551
00
105
11
Overall Survival: P+C vs E, Total Population
Data cutoff date: Jun 13, 2018.
Events HR (95% CI) PPembro + Chemo 70% 0.77
(0.63-0.93)0.0034
EXTREME 80%
Median (95% CI)13.0 mo (10.9-14.7)10.7 mo (9.3-11.7)
12-mo rate53.0%43.9% 24-mo rate
29.0%18.7%
Burtness KN048 ESMO 2018
0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 00
1 0
2 0
3 0
4 0
5 0
6 0
7 0
8 0
9 0
1 0 0
M o n th s
On
go
ing
Re
sp
on
se
, %
N o . a t R is k49 38 32 13 489 34 11 5 0
62
00
259
00
Response Summary, P vs E
aPatients without measurable disease per central review at baseline who did not have CR or PD. bPatients who did not have a post-baseline imaging assessment evaluable for response or who did not have post-baseline imaging. Response assessed per RECIST v1.1 by blinded, independent central radiologic review. Data cutoff date: Jun 13, 2018.
Confirmed Response, n (%)
PembroN = 133
EXTREMEN = 122
ORR 31 (23.3) 44 (36.1)
CR 10 (7.5) 4 (3.3)
PR 21 (15.8) 40 (32.8)
SD 40 (30.1) 42 (34.4)
PD 42 (31.6) 13 (10.7)
Non-CR/non-PDa 8 (6.0) 6 (4.9)
Not evaluable or assessedb 12 (9.0) 17 (13.9)
0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 00
1 0
2 0
3 0
4 0
5 0
6 0
7 0
8 0
9 0
1 0 0
M o n th s
On
go
ing
Re
sp
on
se
, %
N o . a t R is k31 25 21 11 344 16 5 2 0
50
00
173
00
CPS ≥20 CPS ≥1
Duration of response, median (range)P: 20.9 mo (2.7 to 34.8+)E: 4.2 mo (1.2+ to 22.3+)
Confirmed Response, n (%)
PembroN = 257
EXTREMEN = 255
ORR 49 (19.1) 89 (34.9)
CR 14 (5.4) 7 (2.7)
PR 35 (13.6) 82 (32.2)
SD 72 (28.0) 83 (32.5)
PD 100 (38.9) 34 (13.3)
Non-CR/non-PDa 11 (4.3) 11 (4.3)
Not evaluable or assessedb 25 (9.7) 38 (14.9)
Duration of response, median (range)P: 20.9 mo (1.5+ to 34.8+)E: 4.5 mo (1.2+ to 28.6+)
Burtness KN048 ESMO 2018
0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 00
1 0
2 0
3 0
4 0
5 0
6 0
7 0
8 0
9 0
1 0 0
M o n th s
On
go
ing
Re
sp
on
se
, %
N o . a t R is k1 0 0 6 2 2 9 1 3 11 0 1 3 8 1 4 6 0
31
00
2 01 1
00
Response Summary, P+C vs E, Total Population
aPatients without measurable disease per central review at baseline who did not have CR or PD. bPatients who did not have a post-baseline imaging assessment evaluable for response or who did not have post-baseline imaging. Response assessed per RECIST v1.1 by blinded, independent central radiologic review. Data cutoff date: Jun 13, 2018.
Confirmed Response, n (%)
Pembro + ChemoN = 281
EXTREMEN = 278
ORR 100 (35.6) 101 (36.3)
CR 17 (6.0) 8 (2.9)
PR 83 (29.5) 93 (33.5)
SD 78 (27.8) 94 (33.8)
PD 48 (17.1) 34 (12.2)
Non-CR/non-PDa 13 (4.6) 9 (3.2)
Not evaluable or assessedb 42 (14.9) 40 (14.4)
Duration of Response
Median (range)P+C: 6.7 mo (1.6+ to 30.4+)E: 4.3 mo (1.2+ to 27.9+)
Burtness KN048 ESMO 2018
PUTATIVE FUTURE CHANGES IN 1ST LINE RM-SCCHN
CPS≧20
Standard:− Pembrolizumab
PUTATIVE FUTURE CHANGES IN 1ST LINE RM-SCCHN
CPS≧20 CPS <1
Options:− PF-C− Pembro+CTX?
Standard:− Pembrolizumab
PUTATIVE FUTURE CHANGES IN 1ST LINE RM-SCCHN
CPS≧20 CPS <1
Options:− PF-C− Pembro+CTX?
CPS 1-20
Options:− Pembrolizumab?− Pembro+CTX?− PF-C?
Standard:− Pembrolizumab
PUTATIVE FUTURE CHANGES IN 1ST LINE RM-SCCHN
CPS≧20 CPS <1
Options:− PF-C− Pembro+CTX?
CPS 1-20
Options:− Pembrolizumab?− Pembro+CTX?− PF-C?
Standard:− Pembrolizumab
individualchemotherapy
Cross-overtoCPIorCTXtreatment. NodataafterPembro+CTX!2nd line
1st line
CPI IN METASTATIC CANCERS – NOT A HOME-RUNRight patient and type of treatment are key
1Carbone et al. (2017). New England Journal of Medicine, 376(25), 2415–2426. 2Powles T, et al. EAS 2017, IMvigor211. 3Cohen et al. ESMO 2017: LBA 45. 4 Long er al. JCO 36, 2018 (suppl; #108).5 Kwon et al. (2014). Lancet Oncol15(7), 700–712. http://doi.org/10.1016/S1470-2045(14)70189-5. 6Quellenangabe: Merck KGaA / Pfizer
CPI IN METASTATIC CANCERS – NOT A HOME-RUNRight patient and type of treatment are key
Entity Failure to achieve endpoint for Scenario
Lung 1 Nivolumab 1st line vs. ChemoBladder 2 Atezolizumab 2nd line vs. ChemoHNC3 Pembrolizumab 2nd line vs. ChemoMelanoma4 Epacadostat + Pembrolizumab 1st line vs. PembroProstate5 Ipilimumab Placebo
Ovarian6 Avelumab 2nd line (± Caelyx)tbc … …
1Carbone et al. (2017). New England Journal of Medicine, 376(25), 2415–2426. 2Powles T, et al. EAS 2017, IMvigor211. 3Cohen et al. ESMO 2017: LBA 45. 4 Long er al. JCO 36, 2018 (suppl; #108).5 Kwon et al. (2014). Lancet Oncol15(7), 700–712. http://doi.org/10.1016/S1470-2045(14)70189-5. 6Quellenangabe: Merck KGaA / Pfizer
PUTATIVE PREDICTIVE MARKERS
Chen and Mellmann, Nature 2017
Inflamed -PD-L1 -Signature -TMB
PUTATIVE PREDICTIVE MARKERS
Chen and Mellmann, Nature 2017
Anegy -T-cells -MDSC
Inflamed -PD-L1 -Signature -TMB
PUTATIVE PREDICTIVE MARKERS
Chen and Mellmann, Nature 2017
Anegy -T-cells -MDSC
IC-exclusion
Inflamed -PD-L1 -Signature -TMB
TUMOR MUTATIONAL BURDEN (TMB)A putative predictive marker for CPI treatments
Samstein, R. M. et al. Tumor mutational load predicts survival after immunotherapy across multiple cancer types. Nat. Genet. 44, 1069 (2019).
LETTERS NATURE GENETICS
tumors were profiled by next-generation sequencing and who had received at least one dose of ICI therapy, representing a vari-ety of cancer types with a sufficient number of patients for analysis (Supplementary Fig. 1). Patients who had received atezolizumab, avelumab, durvalumab, ipilimumab, nivolumab, pembrolizumab or tremelimumab as monotherapy or in combination were included in the study. Most patients (1,446, representing 94% of tumors exclud-ing glioma) had stage IV or metastatic disease. A small number of patients had locoregionally recurrent disease (n = 10) or were melanoma patients with regionally advanced unresectable disease (stage III, n = 989 (Supplementary Table 1). In total, 146 received anti-CTLA-4, 1,256 received anti-PD-1 or PD-L1, and 260 received a combination of anti-CTLA-4 and anti-PD-1/PD-L1 therapies. A large number of patients had cancers for which ICI is FDA-approved, including 350 NSCLCs, 321 melanomas, 151 renal cell carcinomas, 214 bladder cancers and 138 head and neck squamous cell cancers (Supplementary Table 2). To calculate TMB, the total number of somatic nonsynonymous mutations was normalized to the total number of megabases sequenced. Overall survival was measured from the date of first ICI treatment to time of death or most recent follow-up. The median follow-up was 19 months (range 0–80, with 830 (50%) patients alive and censored at most recent follow-up).
We defined TMB subgroups by percentile within each histology. We took this approach because the median and range of mutational load have been shown to vary across tumor types13; therefore, a universal cutoff for ‘high TMB’ would be enriched for tumor types with higher mutation load. Across the entire cohort, stratifying tumors by TMB decile within histology showed that a higher number of mutations was associated with improved overall survival. This significant association, stratified by histology, was seen across a variety of cutpoints chosen to define the high-TMB group (ranging from the top 10–50%; Fig. 1a and Supplementary Figs. 3 and 4). A clear trend toward decreasing hazard ratio (HR) of death with increasing TMB cutoff was observed across cancer types, demonstrating increasing benefit from ICI with higher TMB (Fig. 1b and Supplementary Fig. 3)14.
To confirm that these results were present across multiple cancer types, we performed two additional analyses. First, a multivariable analysis across the entire cohort using Cox proportional-hazards regression demonstrated that the tumor mutation burden was signif-icantly associated with overall survival both as a continuous variable (HR = 0.985, P = 3.4 × 10−7) and with a binary cutoff (top 20% of each histology, HR = 0.61, P = 1.3 × 10−7), with adjustment for cancer type, age, drug class of ICI and year of ICI start (Table 1). Furthermore, this association remained significant with removal of melanoma and NSCLC patients from the cohort (Supplementary Table 2), thus indi-cating that this effect was not solely driven by these histologies.
We also performed a stratified analysis within each cancer type by selecting the highest mutation load quintile (top 20%) in each histology as the TMB-high group. Using this approach, we observed a similar association of longer overall survival with higher TMB (top 20% within each histology) across multiple cancer types (Fig. 2 and Supplementary Fig. 5). Although the effect for some individual cancers did not reach statistical significance, possibly because of smaller sample size, the numerical trend of better overall survival (HR < 1) was observed in nearly all cancer types, with glioma the clearest exception. Together, these data indicate that the association between TMB and improved survival after ICI is likely to be present in most cancer histologies.
Consistent with varying distributions of TMB across histologies, the TMB cutoff associated with the top 20% of each cancer type varied markedly (Fig. 2). Importantly, this result suggests that there is not likely to be a universal number defining high TMB that is pre-dictive of clinical benefit to ICI across all cancer types, and that the optimal cutpoint is likely to vary for different cancers.
A similar numerical trend was observed for longer overall survival with TMB measured as a continuous variable across many
0 20 40 600.0
0.5
1.0
TMB cutoff
Haz
ard
ratio
a
b
0 12 24 36 480
50
100
Time (m)
Ove
rall
surv
ival
(%)
Top 10% TMB within histology
Top10–20% TMB within histology
P = 1.7 × 10–6
No. at risk
586
100
101
231
39
43
85 33
5
6
1,305
184
173
Bottom 80%
Top10–20%
Top10%
Bottom 80% TMB within histology
16
16
Fig. 1 | Effect of mutational load on overall survival after ICI treatment. a, Kaplan–Meier curves for patients with tumors falling into the depicted deciles of TMB within each histology. Overall survival is from the first dose of ICI. Two-sided log-rank P values are indicated for all patients, with univariate Cox regression HR of 0.76 (95% confidence interval (CI) 0.62–0.94) and 0.52 (95% CI 0.42–0.64) for the 10–20% and top 10% groups, respectively, compared with the bottom 80% group. b, Cox regression hazard ratios for overall survival of 1,662 patients, at the depicted percentile cutoffs of TMB across all cancer subtypes. Solid black circles represent HRs with P!< !0.05 (two-sided log-rank P value).
Table 1 | Multivariable analysis of factors associated with overall survival
HR 95% CI P value
Normalized mutation count Continuous 0.985 0.979–0.991 3.4!× !10−7
Binary (top 20% of each histology)
0.61 0.508–0.733 1.3!× !10−7
Cancer type Melanoma (reference) NSCLC 2.08 1.61–2.68 1.9!× !10−8
Not melanoma/NSCLC 1.52 1.21–1.92 3.7!× !10−4
Age 0.995 0.990–1.004 0.07Drug class PD-1/PD-L1 (reference) CTLA-4 1.18 0.846–1.660 0.32 Combination 0.67 0.534–0.844 6.6!× !10−4
Year of ICI start 2.3!× !10−8
Cox proportional hazards multivariable analysis of overall survival in 1,662 patients treated with immune checkpoint inhibitor (ICI) demonstrating the hazard ratios for individual covariates.
NATURE GENETICS | www.nature.com/naturegenetics
LETTERSNATURE GENETICS
histologies, concordant with the number of patients in the subgroup (Supplementary Fig. 6). In agreement with differences in overall survival, we also observed similar associations between TMB and rates of objective response/clinical benefit to ICI, or progression-free survival, in patients with cancer types for which response data were available—NSCLC, melanoma, esophagogastric, head and neck, and renal cell cancer15–17 (Supplementary Figs. 7 and 8).
To investigate the possibility that the observed survival dif-ferences among patients with higher-TMB tumors might simply be attributable to a general prognostic benefit of high mutational load, unrelated to ICI, we analyzed the outcomes of 5,371 patients with metastatic cancers who did not receive ICI and whose tumors were sequenced with MSK-IMPACT. In these patients, there was no association between higher TMB and improved overall sur-vival (HR = 1.12, P = 0.11). This lack of prognostic benefit was also observed within each histology (Supplementary Figs. 5 and 9).
Of note, the TMB cutpoint for the top 20% of colorectal cancer patients was high (52.2/Mb), potentially consistent with many MSI-high colorectal tumors receiving ICI treatment. To evaluate the pos-sibility that the ICI-treated cohort of patients might be enriched for those with higher TMB (if, for example, clinicians were more likely to triage higher-TMB patients to ICI therapy), we repeated the sur-vival analyses, instead calculating the top 20% of TMB among all (both ICI- and non-ICI-treated) patients. The TMB cutpoints in other cancer types were not changed with this calculation, and the associations with survival in each cancer type remained very simi-lar in both the ICI- and non-ICI-treated cohorts (Supplementary Figs. 10 and 11).
Distinctly from the other cancer types, there was no association between higher TMB and improved survival in patients with glioma; in fact, the trend was toward poorer survival. Although there have been case reports of dramatic responses to ICI in patients with glio-blastoma associated with childhood biallelic mismatch repair defi-ciency18, mismatch repair is very rare in glioblastoma, and higher TMB in many glioma patients may reflect previous exposure to the alkylating agent temozolomide, which can promote the expansion
of less immunogenic subclonal mutations19. Alternatively, antitu-mor immune responses in the central nervous system may be dis-tinct and less dependent on TMB.
As would be expected in a large multicancer analysis of tumors sequenced as part of clinical care, the patients included were het-erogeneous: some had been heavily pretreated, whereas others were treated with a variety of combination therapies. The timing of MSK-IMPACT testing relative to ICI start was also variable. Nevertheless, the finding of a significant association with overall survival in a heterogeneous cohort underscores the robustness of TMB as a predictive biomarker, thus suggesting that it is likely to be clinically meaningful.
TMB, as measured by targeted NGS panels such as MSK-IMPACT, has previously been shown to have a high correlation with total mutational burden, as measured by whole-exome sequenc-ing20–24. MSK-IMPACT offers the advantage of matched normal germline sequencing for each patient, permitting precise identifica-tion of true somatic mutations.
Although TMB measured in exome sequencing is highly cor-related with measurements in targeted sequencing, it is impor-tant to note that numerical cutpoints may differ across platforms. Additionally, we note that TMB cutoffs for individual histologies may not represent the ideal values for clinical use, and they are shown primarily to demonstrate that a relationship exists between TMB and survival for each histology. We chose a top-twentieth per-centile cutoff for TMB to dichotomize our data, but this does not imply any clinical significance to this threshold.
The variable threshold of TMB across histologies can probably be attributed to distinct tumor microenvironments as well as the numerous other factors shown to independently predict response to ICI, including clonality, immune infiltration, immune cell exclu-sion, human leukocyte antigen genotype and alterations and expres-sion levels of checkpoint molecules, as well as others19,25–28. Our data overall suggest that TMB is associated with increasing overall survival in a dose-dependent fashion. The pancancer nature of this biomarker probably reflects fundamental mechanisms by which
Cancer type
All samples in cohort
Bladder
Breast
ER+
ER−
Unknown primary
Colorectal
Esophagogastric
Glioma
Head and neck
Melanoma
Non-small cell lung
Renal cell carcinoma
Combo
CTLA4
PD-1/PDL-1
No. of patients
1,662
214
45
24
21
90
110
126
117
138
321
350
151
260
146
1,256
Cutoff
−
17.6
5.9
6.8
4.4
14.2
52.2
8.8
5.9
10.3
30.7
13.8
5.9
−
−
−
P-value
1.59 × 10–6
0.040
0.605
0.287
0.731
0.155
0.031
0.221
0.465
7.42 × 10–3
0.067
2.30 × 10–4
0.569
0.018
1.89 × 10–3
6.95 × 10–4
0.12 0.25 0.50 1.0 2.0 4.0
<−− Better overall survival--------HR---------Worse overall survival−−>
Drug class
Fig. 2 | Effect of nonsynonymous mutational load on overall survival after ICI treatment, by cancer subtype and drug class. Forest plot for all patients in the identified cohort or individual cancer subtypes. Indicated are the number of patients and HR comparing overall survival after ICI in patients in the highest twentieth-percentile TMB within each histology. Bars represent the 95% CI. The cutoff defining the top 20% of normalized mutational burden from MSK-IMPACT for each cancer type is shown, as well as the two-sided log-rank P value for the comparison of high and low mutational burden survival curves. ER, estrogen receptor. All cancer types in analysis are displayed.
NATURE GENETICS | www.nature.com/naturegenetics
MULTIPLEX BIOMARKER ANALYSESA pattern of markers is more likely to increase probability of response to CPI
Ott, P. A. et al. T-Cell–Inflamed Gene-Expression Profile, Programmed Death Ligand 1 Expression, and Tumor Mutational Burden Predict Efficacy in Patients Treated With Pembrolizumab Across 20 Cancers: KEYNOTE-028. J Clin Oncol 37, 318–327 (2019).
20 Cancers. N=471. Based on KN028 trial
occurs in tumors with high levels of both inflammatory andmutational biomarkers. Of note, expression of PD-L1 IHC byCPS was associated with clinical efficacy, which demon-strates that, even in a PD-L1–selected population (viaprototype assay), the level of PD-L1 expression by CPS inthe tumor provided additional information about expected
clinical outcome across the evaluated cohorts. Expressionof PD-L1 via IHC was moderately correlated with theT-cell–inflamed GEP, as expected for a gene coexpressedwithin the profile and consistent with the known PD-L1upregulation in a T-cell–activated TME.37-39 Similarly, themoderate correlation of both PD-L1 CPS and T-cell–inflamed
Responder Status
T-Ce
ll–In
flam
ed G
EP s
core
–1.0
–0.5
0.0
0.5
Nonresponder Responder
A
T-Cell–Inflamed GEP score
Prog
ress
ion-
Free
Sur
viva
l (da
ys)
0
200
400
600
800
1,000
–1.0 –0.5 0.0 0.5
Progressive disease/nonevaluable
Partial or complete response
Stable disease
B
C
−2 20
Column z score
Color Key
Response
CXCR6TIG
ITCD27
PDCD1LG2 (
PD−L2)
CD274 (
PD−L1)
CD8ALA
G3NKG7
CCL5
CMKLR
1
PSMB10
CXCL9ID
O1
HLA.D
QA1
CD276
STAT1
HLA.D
RB1
HLA.E
FIG 3. Association of T-cell–-inflamedgene-expression profile (GEP) withclinical efficacy in the total study cohort.(A) Relationship between GEP scoreand objective response rate. Responder,complete orpartial response; nonresponder,not complete or partial response. (B) Re-lationship between GEP and progression-free survival in patients with completeor partial responses, stable disease,and progressive disease/nonevaluabledisease. Censoring information not shown.(C) Pattern of expression of the 18 genesin the T-cell–-inflamed GEP and clinicalresponse. Heat map of the 18-geneT-cell–-inflamed GEP. Expression levelshave been standardized (centered andscaled) within columns for visualization.The rows and columns have been groupedusing unsupervised clustering. Black linesin the left response column indicate re-sponders; gray lines, nonresponders.
324 © 2018 by American Society of Clinical Oncology Volume 37, Issue 4
Ott et al
Downloaded from ascopubs.org by DFG Deutsche Forschungsgemeinschaft on March 8, 2019 from 132.252.148.009Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
GEP with TMB, in which each shows predictive value in-dividually, suggests that multiple clinical-grade assays toinform patient treatment may be important for the best se-lection of patients to receive anti–PD-1 monotherapy.
The results of this analysis suggest that all three bio-markers, when assessed separately or in combination,predict clinical efficacy with pembrolizumab across mul-tiple tumor types, consistent with previous data that suggestthat these biomarkers function as complementary pre-dictors of response to pembrolizumab.12-14 The use of jointinformation from these biomarkers may help identify smallprevalence populations within cancer types that have lowORRs for which anti–PD-1 monotherapy may still bebeneficial. Comparisons of the clinical utility of the bio-markers was not deemed appropriate for this data set;larger studies are needed to obtain more precision aboutthe nature of these relationships within individual cancertypes and to assess the potential clinical utility of developingcutoffs for such biomarkers in patient selection for anti–PD-1 therapy. In addition, assessment of these biomarkers in arandomized, comparative setting is required to provide a
better understanding of the predictive versus prognosticelements of these relationships and how different com-ponents of the TME and mutational status may beused to predict outcome with anti–PD-1 monotherapyand combination therapies relative to standard-of-caretreatment.
Although this is a fairly large study that shows significantrelationships among the biomarkers and clinical outcomeacross multiple tumor types, there are limitations. Therequirement of PD-L1 positivity for trial participation mayhave skewed the distribution of the biomarkers evaluated inthis data set compared with those in an all-comers (anylevel of PD-L1 expression) setting. Because of the PD-L1preselection and the small representation from each cancertype, estimation of response rates at biomarker cut pointsand detailed analysis of potential clinical utility are beyondthe scope of this study. Nonetheless, the significant re-lationships demonstrated between the biomarkers and clinicaloutcome are consistent with previous observations.12-14 De-spite differences in scoring and staining characteristicsshown for various PD-L1 assay methods and associated
Responder Status
10
20
50
100
200
500
1,000
2,000
5,000
Nonresponder Responder
A
10
20
50
100
200
500
1,000
2,000
5,000 Progressive disease/nonevaluable
Partial or complete response
Stable disease
T-Cell– Inflamed GEP Score–1.0 –0.5 0.0 0.5
C
B
10 20 50 100
200
500
1,000
2,000
5,000
Prog
ress
ion-
Free
Sur
viva
l (da
ys)
0
200
400
600
800
1,000
Progressive disease/nonevaluable
Partial or complete response
Stable disease
Tum
or M
utat
iona
l Loa
d(lo
g 10 s
cale
)Tu
mor
Mut
atio
nal B
urde
n(lo
g 10 s
cale
)
Tumor Mutational Burden(log10 scale)
FIG 4. Association of tumor mutationalburden with clinical efficacy in the totalstudy cohort. (A) Relationship betweentumor mutational burden and objectiveresponse rate. Responder, complete orpartial response; nonresponder, notcomplete or partial response. (B) Re-lationship between tumor mutationalburden and progression-free survival inpatients with complete or partial re-sponses, stable disease, and pro-gressive disease/nonevaluable disease.Censoring information not shown. (C)Relationship between T-cell–-inflamedgene-expression profile (GEP) and tu-mor mutational burden in patients withcomplete or partial responses, stabledisease, and progressive disease/non-evaluable disease.
Journal of Clinical Oncology 325
Pan-Tumor Biomarkers Predictive of Response to Pembrolizumab
Downloaded from ascopubs.org by DFG Deutsche Forschungsgemeinschaft on March 8, 2019 from 132.252.148.009Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
GEP with TMB, in which each shows predictive value in-dividually, suggests that multiple clinical-grade assays toinform patient treatment may be important for the best se-lection of patients to receive anti–PD-1 monotherapy.
The results of this analysis suggest that all three bio-markers, when assessed separately or in combination,predict clinical efficacy with pembrolizumab across mul-tiple tumor types, consistent with previous data that suggestthat these biomarkers function as complementary pre-dictors of response to pembrolizumab.12-14 The use of jointinformation from these biomarkers may help identify smallprevalence populations within cancer types that have lowORRs for which anti–PD-1 monotherapy may still bebeneficial. Comparisons of the clinical utility of the bio-markers was not deemed appropriate for this data set;larger studies are needed to obtain more precision aboutthe nature of these relationships within individual cancertypes and to assess the potential clinical utility of developingcutoffs for such biomarkers in patient selection for anti–PD-1 therapy. In addition, assessment of these biomarkers in arandomized, comparative setting is required to provide a
better understanding of the predictive versus prognosticelements of these relationships and how different com-ponents of the TME and mutational status may beused to predict outcome with anti–PD-1 monotherapyand combination therapies relative to standard-of-caretreatment.
Although this is a fairly large study that shows significantrelationships among the biomarkers and clinical outcomeacross multiple tumor types, there are limitations. Therequirement of PD-L1 positivity for trial participation mayhave skewed the distribution of the biomarkers evaluated inthis data set compared with those in an all-comers (anylevel of PD-L1 expression) setting. Because of the PD-L1preselection and the small representation from each cancertype, estimation of response rates at biomarker cut pointsand detailed analysis of potential clinical utility are beyondthe scope of this study. Nonetheless, the significant re-lationships demonstrated between the biomarkers and clinicaloutcome are consistent with previous observations.12-14 De-spite differences in scoring and staining characteristicsshown for various PD-L1 assay methods and associated
Responder Status
10
20
50
100
200
500
1,000
2,000
5,000
Nonresponder Responder
A
10
20
50
100
200
500
1,000
2,000
5,000 Progressive disease/nonevaluable
Partial or complete response
Stable disease
T-Cell– Inflamed GEP Score–1.0 –0.5 0.0 0.5
C
B
10 20 50 100
200
500
1,000
2,000
5,000
Prog
ress
ion-
Free
Sur
viva
l (da
ys)
0
200
400
600
800
1,000
Progressive disease/nonevaluable
Partial or complete response
Stable disease
Tum
or M
utat
iona
l Loa
d(lo
g 10 s
cale
)Tu
mor
Mut
atio
nal B
urde
n(lo
g 10 s
cale
)
Tumor Mutational Burden(log10 scale)
FIG 4. Association of tumor mutationalburden with clinical efficacy in the totalstudy cohort. (A) Relationship betweentumor mutational burden and objectiveresponse rate. Responder, complete orpartial response; nonresponder, notcomplete or partial response. (B) Re-lationship between tumor mutationalburden and progression-free survival inpatients with complete or partial re-sponses, stable disease, and pro-gressive disease/nonevaluable disease.Censoring information not shown. (C)Relationship between T-cell–-inflamedgene-expression profile (GEP) and tu-mor mutational burden in patients withcomplete or partial responses, stabledisease, and progressive disease/non-evaluable disease.
Journal of Clinical Oncology 325
Pan-Tumor Biomarkers Predictive of Response to Pembrolizumab
Downloaded from ascopubs.org by DFG Deutsche Forschungsgemeinschaft on March 8, 2019 from 132.252.148.009Copyright © 2019 American Society of Clinical Oncology. All rights reserved.
low association: r=0.29. P=0.007
MULTIPLEX BIOMARKER IN SCCHN
Oliva, M. et al. Immune biomarkers of response to immune-checkpoint inhibitors in head and neck squamous cell carcinoma. Annals of Oncology 30, 57–67 (2019).
significant correlation with treatment efficacy or survival[9, 133]. However, the study had several limitations, includingthe lack of uniformity in sample collection, the small number ofresponses for correlation and importantly, the omission of intes-tinal microbiota. The predictive role of the oral microbiota wasalso investigated in melanoma patients treated with anti-PD-1/PD-L1 therapy, again reporting no association with treatmentoutcome, in contrast to the positive correlation observed with theintestinal microbiota composition [22]. Differential bacterialcomposition between these anatomical sites suggests oral and in-testinal microbiota likely represent distinct entities with specificdisease associations.
Considering the immunomodulatory effects of the intestinalmicrobiota and the growing evidence of the oral microbiotaimpacting HNSCC tumorigenesis and progression, the study oftheir role as a predictive biomarker of response to ICI in this dis-ease is warranted. Hence, our group is currently conducting a re-search study at the Princess Margaret Cancer Centre toprospectively evaluate the oral and intestinal microbiota in ahomogeneous cohort of patients diagnosed with locoregionallyadvanced OPSCC treated with definitive chemoradiotherapy.The overarching goal of this project is to characterize and explorethe correlation with both oral and intestinal microbiota meas-ured in the saliva and stool, respectively, by using 16S rRNAsequencing, in order to obtain a deeper understanding of their
relationship with treatment response. The results of this ongoingstudy will serve as a fundamental basis to evaluate oral and intes-tinal microbiota signatures and their role as predictors of re-sponse to ICI in patients treated within the CCTG HN.9 clinicaltrial, a multicenter phase II noncomparative randomized studyevaluating ICI plus RT followed by maintenance ICI versusstandard chemoradiotherapy in intermediate-risk, HPVþ locore-gionally advanced OPSCC (NCT034106615).
Discussion
Conclusion
Anti-PD-1 agents have become the standard of care for theplatinum-refractory R/M HNSCC. Results from clinical trialsevaluating their role in additional disease settings are pending,but clearly such compounds are already an important therapeuticbackbone in this malignancy. As such, appropriate selection ofpatients who will benefit from these therapies is crucial. To date,there are no validated predictive biomarkers of response that areapplicable uniformly to all HNSCC patients, although many can-didate biomarkers with promising results are undergoing investi-gations. A systematic computational analysis of all clinically
CD8 +
APCAPC
APC HNSCC
TUMOR PROGRESSION
TUMORIGENESIS
Oral mucosae
Tumor antigens
Pro-inflammatorycytokines
IMMUNE UPREGULATION
Oral microbiota
Intestinal microbiota
IMMUNOSUPRESSIVETME
Genomic Lanscape/TMB
CD8 +
CD8 +
PD-L1PD-L1+
-
Macrophages
MDSC
MDSC
ANTI-PD-1/PD-L1
TregsTregs
TregsLAG3+ CD8+PD1+
CD8+
TIM3+ CD8+
LOCAL/SYSTEMIC IMMUNE DOWNREGULATION
HPVHPV
HPV
Figure 1. Interactions between the oral and intestinal microbiome, immune responses and the HNSCC TME. The composition of the oral micro-biota alters the oral mucosae contributing to tumor development and progression in the context of other coexisting factors such as HPV infec-tion. Intestinal and oral microbial composition and diversity regulate systemic and local immune responses modulating the TME along with otherimmune biomarkers such as TMB or immune checkpoint protein expression, ultimately dampening or enhancing antitumor immune responses.
Annals of Oncology Review
Volume 30 | Issue 1 | 2019 doi:10.1093/annonc/mdy507 | 63
Dow
nloaded from https://academ
ic.oup.com/annonc/article-abstract/30/1/57/5194333 by U
niversitaetsbibliothek Duisburg-Essen user on 06 M
ay 2019
SAFETY AND TOLERABILITY
PD-1I ARE SAFE AND TOLERABLE
Ferris et al. ASTRO 2018: LBA10.
Articles
www.thelancet.com Published online November 30, 2018 http://dx.doi.org/10.1016/S0140-6736(18)31999-8 9
183 patients treated with standard of care. In the total population, four patients treated with pembrolizumab and two patients treated with standard of care died from adverse events attributed by the investigator to treatment. The treatment-related events that led to death were death of unspecified cause, large intestine perforation, malignant neoplasm progression, and Stevens-Johnson syndrome in the pembrolizumab group and malignant neoplasm progression and pneumonia in the standard-of-care group. All but one of the deaths in the pembro-lizumab group occurred in patients with a combined positive score of 1 or higher.
The most common treatment-related adverse event was hypothyroidism (33 of 246 patients [13%]) with pembrolizumab and fatigue (43 of 234 patients [18%]) with standard of care (table 2). In the pembrolizumab group, there were four treatment-related adverse events of grade 3–5 severity that occurred in two or more patients each compared with 19 such events in the standard-of-care group. A summary of all treatment-related adverse events is available in the appendix. The adverse events of interest with regard to pembrolizumab, regardless of attribution to treatment by the investigator, are summarised in table 2; one of 246 (<1%) patients had a grade 5 event (ie, resulting in death), which was a severe skin reaction (Stevens-Johnson syndrome).
DiscussionIn the randomised, open-label, phase 3 KEYNOTE-040 trial, pembrolizumab prolonged overall survival com-pared with investigator’s choice of methotrexate, docetaxel, or cetuximab in patients with recurrent or metastatic squamous cell carcinoma of the head and neck. The benefit of pembrolizumab compared with standard-of-care therapy was greater in patients with PD-L1 expression on their tumours or in the tumour microenvironment than in those without PD-L1 expression. Pembrolizumab had a better safety profile than standard of care, with overall profiles consistent with those previously observed and no new or unexpected toxicities. The frequency of adverse events of grade 3–5 severity that were attributed to study treatment by the investigator was 2·7 times lower with pembrolizumab than with standard of care. More patients in the pembrolizumab group died from treatment-related adverse events, although the proportion was low overall (four [2%] of 246 in the pembrolizumab group and two [1%] of 234 in the standard-of-care group).
As previously observed for pembrolizumab and other immune checkpoint inhibitors,8–10,12–14 responses to pembrolizumab were durable. The median duration of response was 18·4 months in the pembrolizumab group, compared with only 5·0 months in the standard-of-care group. Also consistent with previous studies of immune checkpoint inhibitors in the PD-L1–unselected recurrent or metastatic setting was the absence of a progression-free survival benefit for pembrolizumab compared with standard-of-care therapy.10,19–21
In a post-hoc exploratory analysis not adjusted for multiplicity, an interaction between the treatment effect for overall survival and PD-L1 expression appeared to be present, such that benefit of pembrolizumab was greater in patients with a combined positive score of 1 or higher versus those with a combined positive score of less than 1 and those with a tumour proportion score of 50% or higher versus those with a score of less than 50%. Although not formally tested, the benefits on progression-free survival and objective response of pembrolizumab com pared with standard-of-care therapy were greater in patients whose tumours had PD-L1 expression than in those who did not express the ligand. Of note, all four complete responses and 30 of 32 partial responses in the pembrolizumab group occurred in patients with a PD-L1 combined positive score of 1 or higher. Treatment differences were even greater in patients with a PD-L1 tumour proportion score of 50% or higher. The benefit of
Pembrolizumab group (n=246) Standard-of-care group (n=234)
Any grade Grade 3, 4, or 5 Any grade Grade 3, 4, or 5
Treatment-related event*
Any event 155 (63%) 33 (13%) 196 (84%) 85 (36%)
Event leading to treatment discontinuation
15 (6%) 12 (5%) 12 (5%) 9 (4%)
Event leading to death 4 (2%) 4 (2%) 2 (1%) 2 (1%)
Event occurring in 10% or more of patients in either group
Hypothyroidism 33 (13%) 1 (<1%) 2 (1%) 0
Fatigue 31 (13%) 4 (2%) 43 (18%) 2 (1%)
Diarrhoea 20 (8%) 4 (2%) 24 (10%) 1 (<1%)
Rash 19 (8%) 1 (<1%) 34 (15%) 1 (<1%)
Asthenia 18 (7%) 1 (<1%) 28 (12%) 4 (2%)
Anaemia 17 (7%) 1 (<1%) 33 (14%) 9 (4%)
Nausea 12 (5%) 0 29 (12%) 1 (<1%)
Mucosal inflammation 9 (4%) 1 (<1%) 30 (13%) 5 (2%)
Stomatitis 6 (2%) 1 (<1%) 28 (12%) 11 (5%)
Neutrophil count decreased 3 (1%) 1 (<1%) 25 (11%) 20 (9%)
Alopecia 1 (<1%) 0 25 (11%) 0
Event of interest†
Any 63 (26%) 11 (4%) 28 (12%) 11 (5%)
Hypothyroidism 37 (15%) 1 (<1%) 9 (4%) 0
Pneumonitis 10 (4%) 3 (1%) 3 (1%) 3 (1%)
Infusion-related reaction 8 (3%) 1 (<1%) 7 (3%) 1 (<1%)
Severe skin reaction 7 (3%) 4 (2%) 9 (4%) 7 (3%)
Hyperthyroidism 5 (2%) 0 1 (<1%) 0
Colitis 2 (1%) 0 1 (<1%) 0
Guillain-Barré syndrome 2 (1%) 1 (<1%) 0 0
Hepatitis 2 (1%) 1 (<1%) 0 0
The median duration of treatment in this population was 2·8 months (IQR 1·2–6·8) for pembrolizumab, 1·4 months (0·7–2·2) for methotrexate, 1·7 months (1·2–3·9) for docetaxel, and 2·3 months (1·6–5·0) for cetuximab. *Events were attributed to treatment by the investigator and are listed as indicated by the investigator on the case report form and are in descending order of frequency in the pembrolizumab group. †Events of interest are those with an immune-related cause and are considered regardless of attribution to study treatment by the investigator. These events are listed in descending order of frequency in the pembrolizumab group. In addition to the specific preferred terms listed, related terms were also included. Data are number of patients with at least one event (% of patients).
Table 2: Adverse events in the as-treated population
Cohen, E. W. et al. Lancet 1–12 (2018). doi:10.1016/S0140-6736(18)31999-8
TIME TO DETERIORATION FAVOR NIVOLUMAB OVER CHEMO
Harrington et al. (2017). Lancet Oncology, 0(0). http://doi.org/10.1016/S1470-2045(17)30421-7
IMMUNE-RELATED-AES VARY
Champiat et al. (2015). Annals of Oncology, mdv623. http://doi.org/10.1093/annonc/mdv623
Ipilimumab, Nivolumab and Pembrolizumab – rare AEs:
• cardiotoxicity • agranulozytosis • pancytopenia • CNS toxicity
IRAE OCCUR WITH DELAY
WEBER et al. J CLIN ONCOL 2012. Blansfield, J.A. et al. (2005). J. Immunother. 28, 593–598.
Eigentler et al. (2016). Cancer Treatment Reviews, 45, 7–18. http://doi.org/10.1016/j.ctrv.2016.02.003
topic systemic
• creme • timctura • ...
• oral • i.v.
CURRENT COMBINATION TRIALS IN RM-SCCHN
CPI: Checkpunktinhibitor. TKI: Tyrosinkinaseinhibitor. CTX: Chemotherapie. 1Siu et al. MHNCS 2018. 2www.astrazeneca.com/media-centre/press-releases/2018/update-on-the-phase-iii-eagle-trial-of-imfinzi-and-tremelimumab-in-advanced-head-and-neck-cancer-07122018.html. 3Burtness et al LBA ESMO 2018.
*PD-L1low
only
CURRENT COMBINATION TRIALS IN RM-SCCHN
CPI: Checkpunktinhibitor. TKI: Tyrosinkinaseinhibitor. CTX: Chemotherapie. 1Siu et al. MHNCS 2018. 2www.astrazeneca.com/media-centre/press-releases/2018/update-on-the-phase-iii-eagle-trial-of-imfinzi-and-tremelimumab-in-advanced-head-and-neck-cancer-07122018.html. 3Burtness et al LBA ESMO 2018.
*PD-L1low
only
+CPI CPICONDOR*1: DURVA ±TREM vs. TREM EAGLE2: DURVA ±TREM vs. SOC CA209-651: IPI-NIVO vs. PF-C KESTREL: DURVA ±TREM vs. PF-C
✘✘
CURRENT COMBINATION TRIALS IN RM-SCCHN
CPI: Checkpunktinhibitor. TKI: Tyrosinkinaseinhibitor. CTX: Chemotherapie. 1Siu et al. MHNCS 2018. 2www.astrazeneca.com/media-centre/press-releases/2018/update-on-the-phase-iii-eagle-trial-of-imfinzi-and-tremelimumab-in-advanced-head-and-neck-cancer-07122018.html. 3Burtness et al LBA ESMO 2018.
*PD-L1low
only
CPI + CTX KN0483: PEM or PEM-PF vs. PF-C ✓
+CPI CPICONDOR*1: DURVA ±TREM vs. TREM EAGLE2: DURVA ±TREM vs. SOC CA209-651: IPI-NIVO vs. PF-C KESTREL: DURVA ±TREM vs. PF-C
✘✘
CONCLUSIONS
CONCLUSIONS
• Chemotherapy offers limited long-term advantage
CONCLUSIONS
• Chemotherapy offers limited long-term advantage• Immunotherapy adds a new layer of efficacy to R/M-SCCHN
CONCLUSIONS
• Chemotherapy offers limited long-term advantage• Immunotherapy adds a new layer of efficacy to R/M-SCCHN• PD-(L)1 inhibitors have shown clinical efficacy and are part of current
treatment algorithm
CONCLUSIONS
• Chemotherapy offers limited long-term advantage• Immunotherapy adds a new layer of efficacy to R/M-SCCHN• PD-(L)1 inhibitors have shown clinical efficacy and are part of current
treatment algorithm• Role of CTLA-4 + PD-(L)1 inhibition is unclear
CONCLUSIONS
• Chemotherapy offers limited long-term advantage• Immunotherapy adds a new layer of efficacy to R/M-SCCHN• PD-(L)1 inhibitors have shown clinical efficacy and are part of current
treatment algorithm• Role of CTLA-4 + PD-(L)1 inhibition is unclear• Near future changes are ahead in 1st line, introducing single-agent
PD-1 inhibition as treatment approach in PD-L1+ RM-SCCHN
CONCLUSIONS
• Chemotherapy offers limited long-term advantage• Immunotherapy adds a new layer of efficacy to R/M-SCCHN• PD-(L)1 inhibitors have shown clinical efficacy and are part of current
treatment algorithm• Role of CTLA-4 + PD-(L)1 inhibition is unclear• Near future changes are ahead in 1st line, introducing single-agent
PD-1 inhibition as treatment approach in PD-L1+ RM-SCCHN• Selection of patients is key to further clinical development
CONCLUSIONS
• Chemotherapy offers limited long-term advantage• Immunotherapy adds a new layer of efficacy to R/M-SCCHN• PD-(L)1 inhibitors have shown clinical efficacy and are part of current
treatment algorithm• Role of CTLA-4 + PD-(L)1 inhibition is unclear• Near future changes are ahead in 1st line, introducing single-agent
PD-1 inhibition as treatment approach in PD-L1+ RM-SCCHN• Selection of patients is key to further clinical development • PD-L1, signatures and TMB are putative markers