Circulating KRAS G12D but not G12V is associated with survival in metastatic pancreatic ductal adenocarcinoma

Abstract

While high circulating tumor DNA (ctDNA) levels are associated with poor survival for multiple cancers, variant-specific differences in the association of ctDNA levels and survival have not been examined. Here we investigate KRAS ctDNA (ctKRAS) variant-specific associations with overall and progression-free survival (OS/PFS) in first-line metastatic pancreatic ductal adenocarcinoma (mPDAC) for patients receiving chemoimmunotherapy ("PRINCE", NCT03214250), and an independent cohort receiving standard of care (SOC) chemotherapy. For PRINCE, higher baseline plasma levels are associated with worse OS for ctKRAS G12D (log-rank p = 0.0010) but not G12V (p = 0.7101), even with adjustment for clinical covariates. Early, on-therapy clearance of G12D (p = 0.0002), but not G12V (p = 0.4058), strongly associates with OS for PRINCE. Similar results are obtained for the SOC cohort, and for PFS in both cohorts. These results suggest ctKRAS G12D but not G12V as a promising prognostic biomarker for mPDAC and that G12D clearance could also serve as an early biomarker of response.

Fig. 1. CONSORT diagram showing the PRINCE clinical trial and standard of care (SOC) validation cohorts.

Plasma samples were obtained from 200 total patients, including 115 PRINCE (A) and 85 SOC (B) patients. Altogether, 281 samples were analyzed, including: 115 PRINCE samples at baseline (32 for patients with prior therapy plus 83 for patients with no prior therapy), 41 PRINCE samples at week 8 (for patients with no prior therapy), 85 SOC samples at baseline, and 40 SOC samples at week 8. All SOC patients were therapy-naive. Among the 67 PRINCE patients with ctKRAS detected in baseline plasma, 41 also had a week 8 plasma obtained but 26 were unavailable due to the following reasons: 15 discontinued treatment prior to week 8, 6 died prior to week 8, 3 received a dose at week 8 but there was no blood drawn, and 2 did not receive a dose at week 8 due to an adverse event. Among 69 SOC patients with ctKRAS detected in baseline plasma, 40 also had a week 8 plasma obtained but 29 were unavailable due to the following reasons: 8 either discontinued treatment or were lost to follow-up, 13 died prior to week 8, 5 had a follow-up blood draw outside of the week 8 window, and 3 did not have week 8 samples available at the time of analysis. Source data are provided as a Source Data file.

Fig. 2. Survival association for baseline ctKRAS variant allele fraction (VAF) by variant for therapy-naive PRINCE patients.

Shown are the Kaplan–Meier curves for baseline VAF dichotomized at the median for overall survival (top, A, B) and progression-free survival (bottom, C, D) for patients with G12D- (left, A, C) or G12V-bearing tumors (right, B, D). Cox regression hazard ratios (HR) and 95% confidence intervals (CI) are shown with log-rank p-values. Source data are provided as a Source Data file.

Fig. 3. Survival association for baseline ctKRAS variant allele fraction (VAF) by variant for therapy-naive standard of care (SOC) patients.

Shown are the Kaplan–Meier curves for baseline VAF dichotomized at the median for overall survival (top, A, B) and progression-free survival (bottom, C, D) for patients with G12D- (left, A, C) or G12V-bearing tumors (right, B, D). Cox regression hazard ratios (HR) and 95% confidence intervals (CI) are shown with log-rank p-values. Source data are provided as a Source Data file.

Fig. 4. Survival association with early, on-therapy ctKRAS dynamics by variant for PRINCE cohort.

Shown is the association with one-year survival (alive or dead at one year) for therapy-naive PRINCE patients with any detected baseline ctKRAS mutation as measured by A changes in ctKRAS variant allele fraction (VAF) from baseline to week 8 on therapy (n = 37), or B ctKRAS clearance at week 8 on therapy for G12D only (left) and G12V only (right). Shown in C is Kaplan-Meyer analysis dichotomized by ctKRAS clearance vs no clearance and association with overall survival for G12D only (left), and G12V only (right). Results for progression-free survival shown in D. Among the 39 PRINCE patients with both a baseline and week 8 plasma obtained, 2 had insufficient follow-up to determine 1-year survival and were thus excluded from the results shown in A, B. Mann–Whitney test (two-sided) used for comparisons in A. Fishers exact test (two-sided) used in B. Cox regression hazard ratios (HR) and 95% confidence intervals (CI) are shown with log-rank p-values for C, D. Source data are provided as a Source Data file.

Fig. 5. Survival association with early, on-therapy ctKRAS dynamics by variant for SOC cohort.

Shown is the association with 1-year survival (alive or dead at one year) for therapy-naive SOC patients with any detected baseline ctKRAS mutation as measured by A changes in ctKRAS variant allele fraction (VAF) from baseline to week 8 on therapy (n = 31), or B ctKRAS clearance at week 8 on therapy for G12D only (left) and G12V only (right,). Shown in C is Kaplan–Meyer analysis dichotomized by ctKRAS clearance vs no clearance and association with overall survival for G12D only (left), and G12V only (right). Results for progression-free survival shown in D. Mann–Whitney test (two-sided) used for comparisons in A. Fishers exact test (two-sided) used in B. Cox regression hazard ratios (HR) and 95% confidence intervals (CI) are shown with log-rank p values for C, D. Source data are provided as a Source Data file.

Fig. 6. Correlation of clinical variables with baseline ctKRAS variant allele fraction (VAF) levels for therapy-naive PRINCE and standard of care (SOC) patients.

Shown are results for Spearman correlations (two-sided) for all continuous clinical variables with ctKRAS VAF levels for PRINCE patients (top) and SOC patients (bottom). For each variable, the Spearman ρ is shown for patients with ctKRAS G12D (green bars) and G12V (dark gray bars). “Sum of Diameters” and “Count of Lesions” refer to categories of RECIST imaging measurements (see Methods). Among 56 total PRINCE patients analyzed, 33 have ctKRAS G12D and 23 have G12V. For SOC, the number of patients varies slightly by clinical variable (see Supplementary Table 5 for exact patient numbers). Significant p-values are indicated for the bar over which it appears. If only one bar in a pair has a p-value, that indicates significance for the association of the clinical variable with that particular ctKRAS variant (G12D or G12V) VAF levels but not for the other variant. If a p-value appears over both bars in a pair, that indicates that the association of that clinical variant with ctKRAS VAF for both G12D and G12V was significant. ALT alanine transaminase, AST aspartate transaminase, and BUN blood urea nitrogen are used. Source data are provided as a Source Data file.