Background: IRS is a multivariable biomarker algorithm combining tissue-based tumor mutation burden (TMB) with quantitative gene expression biomarkers, previously validated to predict anti-PD-1 and anti-PD-L1 monotherapy (mono) benefit in an observational clinical trial (PMID: 36750617). Herein, we evaluated IRS in pts with advanced solid tumors across the Kaiser Permanente Northern California (KPNC) health system following a pre-specified statistical analysis plan. Methods: KPNC pts with valid IRS results (from 2018 – 2022; specimen collection years 2011-2021) and electronic health record (EHR) documented > 1^st line systemic pembro mono treatment were included; pts with gliomas or previous immunotherapy were excluded. All oncologic treatments were obtained from the EHR and standardized to determine time to next treatment (TTNT) as a real-world progression free survival endpoint. A within-pt evaluation was performed, comparing pembro TTNT to the pt’s immediate prior therapy line (prior) TTNT. The primary objective was to determine if IRS status (-H vs. Low [-L]) predicts pembro clinical benefit in Cox proportional hazards models through testing for interaction between IRS status and pembro/prior therapy TTNT. Results: With a data cutoff of 06/15/2022, 211 KPNC pts (142 2^nd line and 69 > 2^nd line pembro) were eligible and included (from 24 solid tumor types). The median (m) pembro follow-up was 16.0 months. IRS-H pts (n = 77, 36%) had significantly longer pembro vs. prior TTNT (mTTNT 20.3 mo vs. 4.5 mo, log-rank p-value < 0.0001), whereas IRS-L pts (n = 134, 64%) did not (mTTNT 3.3 mo vs. 5.2 mo, log-rank p-value 0.75); the interaction between IRS status and pembro/prior treatment was significant (interaction test p-value < 0.0001). Conclusions: IRS predicted pan-solid tumor pembro mono benefit compared to prior therapy and may be a potential marker for pembro mono treatment decisions. Clinical trial information: NCT03061305.