Supplemental material

Supplemental material

Supp Fig 1. Flow diagram outlining patient cohorts in this study. Dotted line represents that the 25

pembrolizumab monotherapy samples were included in both sets.

Pre-treatment and on-treatment timepoints available

n=47 Cohort 1 (Figure 1) PD-1or PD-L1 inhibitor monotherapy

Cohort 2 (Figure 2) Pembrolizumab monotherapy or chemo-immunotherapy Pembrolizumab monotherapy n=25

Pre-treatment and three-week on-treatment timepoint available

Chemo-immunotherapy

n=16 IL-6 discovery set

Cytokine screening set (Supp Figures 4 and 5)

Clear clinical responses at first scan n=12 irAEs (Figures 3A and 3B) n=2 irAEs (Figure 3C) n=1 CRP, other cytokines, and irAE monitoring set

2

Supp Fig 2. Lower and upper limits of quantitation (vertical lines) were derived from calibrators

and sample concentrations with less than 20% coefficient of variation (horizontal line). The fraction of plasma samples (including both pre- and on-treatment samples) is this reportable measuring range are reported in the top right quadrant. All cytokine levels below the lower limit of quantitation (LLOQ) or above the upper limit of quantiation (ULOQ) were set to the respective limit for the downstream analysis.

0.01 1 10 1000 LL OQ 0 .1 UL OQ 1 00 0.0 0.1 0.2 0.3 0.4 0.5 Mean Conc. C V IL-6 0.01 1 100 1000 LL OQ 0 .1 ULO Q 1 0 0.0 0.1 0.2 0.3 0.4 0.5 Mean Conc. C V CXCL10 0.01 0.1 1 10 100 1000 LLOQ 0.0 5 ULOQ 32. 5 0.0 0.1 0.2 0.3 0.4 0.5 Mean Conc. C V IL-1b 0.01 1 10 1000 LLO Q 0 .1 UL OQ 1 00 0.0 0.1 0.2 0.3 0.4 0.5 Mean Conc. C V IL-10 0.01 0.1 1 10 100 1000 LLOQ 0.0 8 ULO Q 5 0 0.0 0.1 0.2 0.3 0.4 0.5 Mean Conc. C V IL-15 0.01 0.1 1 10 100 1000 0.0 0.1 0.2 0.3 0.4 0.5 Mean Conc. C V IL-17a

Fractionofmeasuredsamplesin

quantifiablerange 295/302 (98%) 306/306 (100%) 152/300 (51%) 303/303 (100%) 300/302 (99%) 256/293 (87%) doi: 10.1136/jitc-2020-000678 :e000678. 8 2020; J Immunother Cancer , et al. Keegan A

Supp Fig 3. Outcomes by IL-6, PD-L1, and TMB levels. A and B. PFS by IL-6 levels at either pre-(A) or on-treatment (B) according to quartile (Q1 lowest to Q4 highest). p=0.87 and p=0.05 comparing all quartiles by logrank test for trend. C and D. PFS by levels of PD-L1 IHC (C) and TMB (D). p=0.92 and p=0.12 comparing all levels by logrank test for trend. N Median PFS (95% CI) Q1 12 20.4 months (4.2 -NR) Q2 12 5.5 months (3.9-NR) Q3 13 4.3 months (3.2-NR) Q4 10 3.3 months (1.8-NR) P = 0.05 N Median PFS (95% CI) Q1 12 5.6 months (3.9-NR) Q2 12 5.5 months (2.0-NR) Q3 12 5.5 months (3.4-NR) Q4 11 4.3 months (3.3-NR) P = 0.87

D

B

C

_{N Median PFS (95% CI)} 0 to <10 mut/Mb 16 4.2 months (3.4-NR) >10 to <20 mut/Mb 7 3.8 months (3.2-NR) >20 mut/Mb 4 33.1 months (7.7-NR) P = 0.12

A

PFS according to pre-treatment IL-6 PFS according to on-treatment IL-6

PFS according to PD-L1 IHC PFS according to TMB levels

N Median PFS (95% CI) PD-L1 <1% 3 7.2 months (3.5-NR) PD-L1 1-49% 4 10.1 months (3.4-NR) PD-L1 >/=50% 22 6.8 months (3.7-NR)

4

Supp Fig 4. Cytokine changes in 12 pembrolizumab monotherapy patients with early radiologic

responses: clear decrease for responders or increase for nonresponders in tumor burden at first scan and on-treatment timepoints three weeks after first dose of therapy. A. Table indicates patient outcomes across the response categories. B. Absolute concentrations of these cytokines from pre-treatment to three weeks after one cycle of pembrolizumab. Each line represents one patient with responders in green and non-responders in red. The dotted horizontal line is the Simoa lower limit of quantitation (as shown in Supp Fig 1). The gray area represents the range of lower quantitation limits of other conventional immunoassays. C. Percentage change in the respective cytokine from pre-treatment to three weeks after first dose of therapy. Results from two independent experiments with different aliquots of the same plasma specimens are shown. Responder (R) Nonresponder (NR). **p<0.01, *p<0.05, ns not significant by Mann Whitney test.

NR R NR R NR R NR R NR R NR R -200 -100 0 100 200 300 400 500 600 % C h a n g e

Signal-finding experiment with 12-patient subgroup ** IL-6 IL-10 CXCL10 ns * IL-15 * IL-17A ** IL-1b ns NR R NR R NR R NR R NR R NR R -200 -100 0 100 200 300 400 500 600 % C h a n g e

Independent experiment with 12-patient subgr oup

* IL-6 IL-10 CXCL10 ns * IL-15 ns IL-17A * IL-1b ns

A

B

C

Pre-C 1 Pre-C 2 0.1 1 10 100 p g /m l IL-6 LLOQ Pre-C 1 Pre-C 2 0.1 1 10 100 p g /m l IL-10 Pr e-C1 Pr e-C2 0.1 1 10 100 1000 p g /m l CXCL10 Pre-C 1 Pre-C 2 0.1 1 10 100 p g /m l IL-17A Responder (R) Non-responder (NR) Pre-C 1 Pre-C 2 0.1 1 10 p g /m l IL-15 Pre-C 1 Pre-C 2 0.1 1 10 100 1000 p g /m l IL-1b Responders Non-responders Total number of patients 6 6 PFS months, range in months 1 to 4 8 to 22 Best Overall Response

PR 5 SD 1 4 PD 2 doi: 10.1136/jitc-2020-000678 :e000678. 8 2020; J Immunother Cancer , et al. Keegan A

Supp Fig 5. Cytokine changes in twelve pembrolizumab monotherapy patients with early

radiologic responses: clear increase for responders or decrease for nonresponders in tumor burden at first scan and on-treatment timepoints three weeks after first dose of therapy. As in Supp Fig 3, except these assays were not carried forward to testing across the entire patient cohort. *p<0.05, ns not significant by Mann Whitney test.

Pr e-C1 Pr e-C2 10 100 p g /m l R _NR -200 -100 0 100 200 300 400 % C h a n g e ns

IL-6R

a

P re-C1 Pre -C2 0.1 1 10 100 p g /m l R NR -200 -100 0 100 200 300 400 % C h a n g e ns

TNF

a

R NR -200 -100 0 100 200 300 400 % C h a n g e * Pre -C1 Pre -C2 0.1 1 10 p g /m l

IL-2

Pre-C 1 Pre-C 2 1 10 100 p g /m l R _NR -200 -100 0 100 200 300 400 % C h a n g e ns

IL-8

Pre -C1 Pre -C2 1 10 100 1000 10000 p g /m l R NR -200 -100 0 100 200 300 400 % C h a n g e ns

CXCL13

IFN

g

R NR -200 0 200 400 600 800 % C h a n g e ns Pre -C1 Pre -C2 0.1 1 10 100 1000 p g /m l Responder (R) Non-responder (NR)

6

Supp Fig 6. Cytokine changes in relation to a severe infectious complication. The plots reflect

concentrations of six cytokines (color key, top right) measured at three-week intervals, prior to each pembrolizumab infusion. Light gray bar reflects duration of pembrolizumab treatment, dark gray bar reflects liver function test elevation measured at onset of abdominal pain, and black bar reflects steroid treatment. Radiologic responses and imaging timepoints are indicated.

26-S ep-2 017 17-O ct-2 017 7-N ov-2 017 28-N ov-2 017 19-D ec-2 017 9-Ja n-20 18 30-J an-2 018 20-F eb-2 018 13-M ar-2 018 3-A pr-2 018 24-A pr-2 018 15-M ay-2 018 5-Ju n-20 18 26-J un-2 018 17-J ul-2 018 7-A ug-2 018 28-A ug-2 018 18-S ep-2 018 9-O ct-2 018 30-O ct-2 018 20-N ov-2 018 0 5 10 15 20 C o n ce n tr a ti o n ( p g /m l) Patient 696, Cholecystitis Pembrolizumab Steroids LFT elevation CT scan PR Flu vaccine _* SD SD SD IL-6 IL-10 IL-17 IL-15 CXCL10 IL1B doi: 10.1136/jitc-2020-000678 :e000678. 8 2020; J Immunother Cancer , et al. Keegan A

Supp Table 1. List of capture and detection antibodies and protein standards used for all Simoa assays.

Supp Table 2.Simoa assay specifications for all Simoa assays including number of steps,

incubation times, enzyme and detection antibody concentrations, sample dilution factors, and reagent volumes.