Abstract Background: Premalignant CH pattern for myeloid malignancies is well established but the characteristics and pattern, if any, that predict B-lymphoid and plasma cell neoplasms is not well understood. Recent WGS data suggest the presence of lymphoid CH as a seperate entity and have linked myeloid CH to inferior outcomes in some pts with multiple myeloma (MM) and diffuse large B cell neoplasms (DLBCL). To delineate this biology, we studied the association between antecedent CH and risk of common B lymphoid neoplasms, including chronic lymphocytic leukemia (CLL), MM and DLBCL in 1200 samples from the Women's Health Initiative (WHI) cohort, which followed participants aged 50-79 for an average of 20 years. Methods: Cases (n=600, 200 each of CLL, MM, DLBCL) included participants with adjudicated diagnoses and baseline blood samples collected prior to disease onset, all with normal WBC counts to avoid early detection bias. Cancer free controls (n=600) were age and follow-up matched using a time-forward approach. Targeted deep sequencing (2000x) of 146 lymphoid and myeloid genes was performed. Variants with ≥1% VAF were classified as pathogenic (P) or variants of uncertain significance (VUS). Baseline characteristics were compared using Chi-squared and Wilcoxon tests. Logistic regression estimated adjusted odds of disease with clonal hematopoiesis (CH); Firth logistic regression assessed mutation-specific risk. Separate models were run for P mutations and with high-impact VUS in exploratory analyses. Two-sided p-values 0.05 were considered significant. Results: The median age (at blood draw) and time to diagnosis of CLL/MM/DLBCL cases was appx 63-64 yrs and 12.4/13.7/12.4 yrs. CBC among cases vs. controls were not clinically different (e.g. CLL cases vs controls: WBC 5.9 vs.5.5, Hb 13.5 vs. 13.4, Plt 235 vs. 244) The top P mutations (muts) among CLL cases were, DNMT3A, KMT2C, TET2, KHL6, PCLP, IGLL5, TP53, STAG2, FAT1 ; muts were seen in 39% cases with 8% showing clonally complex muts (= 2 mutations) vs. 23% in controls. Top pathogenic muts in controls were DNMT3A, KMT2C, TET2, CBL, TP53, FAT1, ARID1B, FAT2. In cases, other than DNMT3A, majority of the muts were subclonal with VAF 10% except for IGLL5 which was present in 2 cases with VAF 10%, suggesting that we indeed captured the premalignant state. Presence of a pCH was associated with increased odds of future CLL (aOR 2.42, 95% C.I, 3.47, p 0.001). Lymphoid only CH had the highest risk (aOR 3.08; 1.83-5.16, p0.001) while myeloid CH had moderate risk (aOR 1.68; 1.08-2.57, p=0.01) Individually, BIRC6 (aOR 7.25, p=0.03), IGLL5 (aOR 22.5, p=0.007), KLHL6 (aOR 7, p=0.03) and SRCAP (aOR 7.35, p=0.03) were associated with CLL risk. When VUS muts were added to the model as exploratory, only NOTCH1/2 (aOR 13/7.17, p0.001) and CHD2 (aOR 27.5, p=0.002) muts retained significance for CLL risk. The top pathologic muts among MM cases were, DNMT3A, KMT2C, RPS15, TET2, ARID1B, JAK2 and NOTCH1; muts were seen in 26.5% cases with 4% showing clonal complexity. While the presence of all CH combined was not associated with increased odds of MM (27% vs. 23 %, aOR 1.19, 95% C.I. 0.81-1.73, p=0.4), specifically RPS15 was linked with increased risk (aOR 2.51, 95% C.I 2.59-121.01, p=0.001). Among VUS muts, RP1L1(aOR 3.01, p=0.004), PCLO (aOR 2.91, p=0.01) and PTPRD (small n=4) retained significance for increased risk. The top pathologic muts among DLBCL cases were, KMT2C, DNMT3A, ASXL1, CREBBP, JAK2, TET2, and GNB1; muts were seen in 27.3% cases with 4.5 % having clonally complexity. CH was not associated with DLBCL (aOR 1.35, 95% C.I 0.85, 2.16, p=0.2). Individually only CREBBP was found to be exclusively mutated in DLBCL cases (small n=4). None of the VUS muts were significant. Conclusions: This is the first large-scale, targeted, deep-sequencing analysis of antecedent lymphoid and myeloid CH 10 yrs. before diagnosis of non-myeloid hematological cancers. CH was associated with increased odds of developing CLL, but not MM or DLBCL. In CLL, the strongest associations were between lymphoid-only CHIP; individual mutations in BIRC6, IGLL5, KLHL6 and SRCAP; and NOTCH and CHD2 VUS. Notably, increased risk was demonstrated even with subclonal mutations and normal WBC. RPS15 mutations were a significant risk factor for MM. These data identify the pre-malignant mutational landscape of CLL and MM and provide valuable insights regarding disease ontogeny and potential early interception strategies.
Desai et al. (Mon,) studied this question.