To the Editor:
In the January 1, 1997 issue of Blood, Sahota et al1 report their findings on the analysis of Ig variable light (VL ) and heavy chain (VH ) gene sequences expressed by multiple myeloma (MM) B-cell clones. The results of this study confirm and extend our own findings at least regarding Vκ genes in 11 MM cases reported recently.2 However, similarities as well as important differences between these two reports can be identified. First, regarding Vκ gene usage by MM cells, 4/9 (44%) MM clones were found to bear an O8-18 Vκ gene.1 Similarly, O8-18 Vκ gene usage was identified in 2/3 (67%) MM clones in the study of Cannell et al3 and in 3/7 (43%) cases by Wagner et al,4 with 1/4 (25%) cases in the latter report carrying the O8-18 gene if functional in-frame Vκ genes were concerned. In our study,2 we identified only one single case of O8-18 Vκ gene usage out of 11 (9%). Taken together the above data indicate that the O8-18 gene is rearranged in about one-third of cases (33%) of MM B cells, raising the possibility of preferential usage in MM. The O2-12 Vκ gene was found to be used in 2/11 (18%) cases in our analysis and by 1/4 (25%) functional Vκ genes in the study of Wagner et al,4 but in no case in the study of Sahota et al.1 Both O8-18 and O2-12 Vκ genes belong to the VκI gene family, whose members have been reported to be used more frequently than others, but with similar frequencies to normal peripheral blood, fetal bone marrow, and lymphoproliferative disorder B cells.5 The solitary Vκ-IV gene family gene was found to be used by 2/11 (18%) cases in our report2 and by 1/4 (25%) cases in the study of Wagner, yielding an overall 20% frequency when the above data are combined. This finding represents an overutilization of the Vκ-IV gene and is compatible with the pattern reported in other B-lymphoproliferative disorders,3 adult peripheral blood, and fetal bone marrow.5 The Vκ-IV family gene was not identified in any of the cases in the study of Sahota et al.1 In contrast, a strikingly high frequency has been observed regarding usage of the A27 (Humκv325) Vκ gene, with 3/9 (33%) cases bearing this particular rearrangement. It is noteworthy that this gene has been found rearranged in no cases in our analysis and that of Wagner et al4 or Cannell et al.3 Humκv325 has been described to rearrange with a very high frequency in B-cell chronic lymphocytic leukemia (B-CLL) cells irrespectively of whether it is expressed or not (abortively rearranged in λ-bearing CLLs)6 and is believed to represent a Vκ gene derived from the autoimmune repertoire of CD5+ B cells expressing Igs with rheumatoid factor activity. Analysis of Vλ genes identified equal distribution between VλI (DPL2, DPL3), VλII (DPL11, HSLV2046) and VλIII (IGLV3S2, DPL23) families. We have found a similar pattern of Vλ gene family usage, as follows: VλI (DPL1, DPL4), VλII (DPL12), VλIII (DPL16, DPL23), and VλVI (IGLV6S1) (Kosmas et al, unpublished observations, March 1997).
Mutational analysis indicated that 4/15 (27%) clonogenic VL genes had significant clustering of mutations in the complementarity determining regions (CDRs) implying a role for antigen selection according to the methodology proposed by Chang and Casali.7 Despite high levels of replacement mutations and R/S ratios in CDRs, in many cases their clustering in CDRs was not at levels of significance (case 1, 5, 6; Tables 3 and 5). It is possible that the application of strict mathematical rules and calculations would lead to an underestimate or a minimum estimate of the importance of antigen selection. Moreover, critical amino acid substitutions within the CDRs, even at low levels, may confer high affinity for antigen and therefore provide the particular B-cell clone with the appropriate advantage for survival after antigen binding. Reanalysis of our data2 according to the modification of the binomial probability model of Chang and Casali indicates that 7/11 (63%) MM clones bear mutations suggestive of antigen selection; a frequency still much higher than that reported by Sahota et al.1 Similar analysis of the data of Wagner et al4 indicates significant clustering of mutations in 2/4 (50%) cases, with the remaining two cases exhibiting a trend to significance (P = .08).
An intriguing observation raised in the current report1 pointed out that comparison of VH and VL sequences derived from the same MM clone yielded significant clustering of mutations for antigen selection either in the VH or in the VL gene, but never in both. These findings should be regarded with caution, since, as the authors indicate, there is no physiological analogue to this phenomenon at least when normal antigen-selected B cells are concerned. A similar analysis of VH and Vκ genes in follicular lymphoma revealed the presence of mutations predominantly in VH genes and to a lesser extent in their partner Vκ genes, with less significant clustering of mutations suggestive of antigen selection in Vκ.8 However, in 2/10 cases, significant clustering of mutations was seen in Vκ genes but not in their corresponding VH genes, thus supporting the suggestion raised,1 that a complementary imprint of antigen selection witnessed by VH and Vκ tumor-derived sequences might be an important event during B-cell ontogeny.
Response
Analysis of V-genes used by human B-cell tumors is providing information that could have relevance for the origin and clonal history of these tumors. Extensive investigation of VH has already shown clear asymmetric gene usage in certain tumor categories,1-1 and some suggestion of asymmetry in MM.1-2 One difficulty is that apparent asymmetry requires careful comparison with the normal expressed repertoire, and this has been difficult due to fluctuation in the available repertoire,1-3 and to antigenic experience.1-4 Information on VL is more limited, and in our view, we are not in a position to comment on asymmetry in usage yet. The figures quoted by Kosmas et al on incidence of the relatively commonly used Vκ gene, O8-18, in cases of multiple myeloma (MM), bear this out in showing a very wide range among various small studies. As more sequences become available, it should be possible to make deductions about possible bias in both Vκ and Vλ gene usage.
Detailed investigations of the Vκ5 and Vλ6 germ line repertoires allow us to comment on the level and nature of somatic mutations in tumor sequences. Location of replacement mutations in the complementarity determining regions (CDRs) can indicate antigen selection, and we have found Chang and Casali's1-7 treatment of sequence data helpful in revealing statistically significant cluster patterns. Our observation that clustering in MM sequences was in either VH or VL was surprising, and we ourselves pointed to the need to compare the patterns in normal B cells. However, it has since been confirmed in cases of Burkitt's lymphoma,1-8 as well as by data from Kosmas et al. It remains an intriguing observation, and our point was that we should not rely solely on analysis of VH to draw conclusions about antigen selection, a lesson which is obvious when we look at models of the antigen-combining site.
