RS occurs in 5–15% CLL, depending on follow-up length and re-biopsy policy. The value of biological and clinical risk factors in predicting RS is unknown and represents the aim of this study. The analysis was based on a consecutive series of 185 CLL. Diagnosis of RS was based on lymph node or extranodal tissue biopsy. RS was diagnosed in 17 cases all represented by diffuse large B-cell lymphoma (DLBCL). Cumulative incidence of RS at 5 and 10 years was 13.6% (95%CI: 7.0–20.1%) and 16.2% (95%CI: 8.0–24.4%), respectively. Transformation plateaued at 82 months of follow-up. Median time to transformation was 23.0 months (95%CI: 15.9–30.1 months) from CLL diagnosis. IGHV studies documented clonal relationship between CLL and DLBCL in 15/17 (88.2%) cases. At the time of CLL diagnosis, molecular variables predicting RS were: IGHV4–39 usage (5-year risk IGHV4-39: 56.2% vs IGHVnon-VH4-39: 11.2%; p<0.001), IGHV homology ≥98% (5-year risk IGHV homology ≥98%: 28.3% vs IGHV homology <98%: 7.0%; p=0.006), absence of del13q14 (5-year risk non-del13q14: 23.1% vs del13q14: 3.8%; p=0.006), and presence of +12 (5-year risk +12: 19.5% vs non +12 10.0%; p=0.027). CLL with p53 inactivation by p53 mutation and/or del17p13 had a trend toward an increased risk of transformation (5-year risk del17p13/p53 mutation: 24.6% vs non-del17p13/p53 mutation: 12.1%; p=0.061). Cluster analysis revealed that 6/17 (35.2%) CLL evolving to RS carried stereotyped CDR3s. Utilization of stereotyped IGHV4-39 or stereotyped IGHV1-2/1-3 predicted transformation (5-year risk stereotyped CDR3: 63.0%; p<0.001). Del11q22-q23, normal FISH katyotype, BCL2 C938A polymorphism, and telomere length did not predict RS. Among phenotypic markers at CLL diagnosis, expression of both CD38 (5-year risk CD38+: 25.4% vs CD38−: 4.7%; p<0.001) and ZAP70 (5-year risk ZAP70+ 27.9% vs ZAP70− 6.7%; p=0.002) predicted RS. At the time of CLL diagnosis, clinical variables predicting RS included: lymph node size ≥3 cm (5-year risk lymph node ≥3 cm: 49.1% vs lymph node <3 cm: 6.1%; p<0.001), involvement of 3 nodal areas (5-year risk 3 nodal areas: 31.8% vs nodal areas <2: 8.3%; p<0.001), Binet B stage (5-year risk Binet B: 40.8% vs Binet C 20% vs Binet A: 7.1%; p=0.003), and LDH ≥1.2 × ULN (5-year risk LDH >1.2 × ULN: 37.5% vs LDH <1.2 × ULN: 8.6%; p=0.001). Age, sex, Rai stage, splenomegaly, Hb, platelet count, percentage of BM lymphocytes, B2M, ALP, albumin and lymphocyte doubling time did not predict RS. Landmark analysis at 12, 24, 36, 48 and 60 months documented:

  • a significantly increased risk of subsequent RS in progressive CLL vs stable CLL starting from 24 months; and

  • no RS in stable CLL after 48 months of follow up.

Multivariate analysis identified CD38 expression (HR 4.22, 95%CI 1.26–14.05; p= 0.019) and IGHV4-39 usage (HR 4.29, 95%CI 1.28–14.29; p=0.018) as biological predictors of RS, and lymph node ≥3 cm as clinical predictor of RS (HR 8.73, 95%CI 2.45–31.05; p=0.001). Our data suggest that RS is predicted by:

  • predominant nodal disease at diagnosis;

  • CD38 expression; and

  • usage of specific IGHV genes, namely IGHV4-39.

These results suggest the need for a close monitoring and a careful biopsy policy in patients with molecular, phenotypic and clinical predictors of RS identified at the time of CLL diagnosis.

Author notes

Disclosure: No relevant conflicts of interest to declare.