Chronic lymphocytic leukemia (CLL) is recognized as a heterogeneous disease.1 Although the clinical staging systems (Rai et al2 and Binet et al3) provide useful methods of assessing patient prognosis, they do not provide an understanding of the basis of the heterogeneity. CLL cells are monoclonal B cells expressing low levels of surface immunoglobulin and, based on the expression of CD5, appear to phenotypically correspond to mantle-zone naive B cells. Because such cells normally do not pass through the lymphoid follicle during maturation, it would be expected that the V region genes that make up the immunoglobulin they express would not be mutated. But Schroeder and Dighiero4 showed that the immunoglobulin VH genes expressed in about half of a series of 75 CLL patients contained mutations as if they had matured in a lymphoid follicle.
Recent work from Hamblin et al5 and Damle et al6 has shown that the presence or absence of immunoglobulin V gene mutations predicts the natural history of the disease. They found that the presence of unmutated V genes was associated with a much poorer prognosis, even within the group of patients in Binet stage A. Importantly, Damle et al showed a strong correlation between expression of CD38 and the presence of unmutated V genes, suggesting that surface phenotyping might be able to identify patients with a poorer prognosis without the requirement for immunoglobulin gene sequencing. We have analyzed the immunoglobulin V genes in 39 of our patients (25 with familial CLL) to investigate this question further.
DNA was extracted from peripheral blood lymphocytes and subjected to polymerase chain reaction (PCR) amplification using primers for the VH and VL genes, as previously reported.7,8 The amplified fragment was ligated into pMOS Blue T-vector and transformed into MOS Blue competent cells (Amersham Life Science, Buckinghamshire, England). Recombinant plasmids were purified from transformed bacteria and selected by restriction analysis. Nucleotide sequencing was performed as previously reported.8 Inserts were sequenced in 2 directions and from multiple independent clones. Sequences were compared with GenBank-EMBL and V base databases.
The mean age of the patients was 62 years (range, 35-95), and the male-female ratio was 0.95; 29 patients were in stage A, 7 in stage B, and 3 in stage C. Twelve (31%) patients expressed unmutated immunoglobulin; details of the the gene rearrangements are shown in Table 1. Greater than 98% homology with the germline sequence was required to be designated as unmutated. Polymorphisms in the V region genes can account for this degree of disparity. Twenty-seven patients (69%) expressed mutated immunoglobulin; details of the gene rearrangements and mutations are shown in Table 2. Twenty-two of the 27 patients with immunoglobulin gene mutations displayed greater than 5% difference compared to their germline counterpart. Further analysis was performed to assess whether the high mutation rate was due to antigen-driven selection.
In the absence of antigen-driven selection, replacement (mutations that change the protein sequence) and silent mutations (those that do not change protein sequence) are typically distributed randomly throughout the protein sequence. A lower than expected frequency of replacement mutations is circumstantial evidence of pressure to maintain the protein sequence due to essential function. By contrast, if the number of replacement mutations exceeds that expected by chance alone, a positive selection process can be inferred. In immunoglobulin genes, clustering of replacement mutations within the complementarity determining regions (CDRs), which form the antigen contact points, is the product of positive selection in the lymphoid follicle. Given the lengths of the CDRs and the framework regions of the molecule, randomly distributed mutations would result in a CDR–framework mutation ratio of less than 0.3. In this series, 20 of 22 cases with greater than 5% mutation rate had a CDR–framework mutation ratio of at least 0.3.
A second test of directed versus random mutation is the ratio of replacement mutations to silent mutations. Random mutation processes produce a ratio of 2.9 of replacement (R) to silent (S) mutations (mutations in the third nucleotide of a codon are often silent). In this series, 14 of the 22 cases with a mutation rate greater than 5% exhibited an R-S ratio of 3 or greater in their CDRs. Nine of these had R-S ratios of less than 2.9 in their framework regions, strong evidence for selective pressure on the CDRs.
Thus, our results confirm previous reports suggesting that at least 2 morphologically indistinguishable forms of CLL exist, one expressing unmutated V genes, as expected based on phenotype, and one expressing mutated V genes in spite of a cell-surface phenotype associated with antigen-naive B cells.
The median followup time was 7 years overall (range, 2-21); 5 years (range, 2-11) for the unmutated group, and 8 years (range, 3-21) for the mutated group. The male-female ratio was not significantly different (P = .2) between the 2 groups: 1.4 in the unmutated group and 0.8 in the mutated group. Most stage A patients (24 of 29) expressed mutated VH genes; most stage B and C patients expressed unmutated VH genes (7 of 10) (P = .036). Among the 12 patients expressing unmutated VH genes, 8 died of CLL-related causes, compared to 6 of 27 patients with mutated VH genes dying, 2 from CLL-related causes and 4 from unrelated causes. Overall (P = .0007) and CLL-specific survival (no more than 10−4) was shorter in patients with unmutated V genes. Among stage A patients, all 5 with unmutated V genes experienced disease progression within 5 years, compared to 7 of 22 evaluable patients in the group with mutated V genes (P = .013). CLL-specific survival was also shorter in stage A patients with unmutated V genes (3 of 5 dead) than in stage A patients with mutated V genes (2 of 24 dead) (P = .0098). We did not examine CD38 expression on the tumor cells.
Thus, as previously reported, CLL exists in at least 2 forms, those with mutated V genes and those with unmutated V genes. Patients with unmutated V genes are more likely to present with advanced disease, more likely to progress clinically, and more likely to die of CLL than those with mutated V genes.
