No association between the presence of killer-cell immunoglobulin-like receptor genes and susceptibility to childhood ALL

To the editor:

Killer-cell immunoglobulin-like receptors (KIRs) play a pivotal role in immunosurveillance and reduction of relapse after hematopoietic stem cell transplantation in acute leukemia.^1,2  Outside the setting of hematopoietic stem cell transplantation, it was recently reported by Almalte et al that inherited KIR genes might be associated with susceptibility or resistance to childhood acute lymphoblastic leukemia (ALL).³  They determined 6 activating KIRs (KIR2DS1-5 and KIR3DS1) in a cohort of 145 children with B-cell leukemia (B-ALL) and 30 children with T-cell leukemia (T-ALL) and compared it with a healthy control group of 245 children of French-Canadian ethnicity. Children with more activating KIR genes had a lower risk for ALL.³  A subsequent study by Babor et al analyzed the presence of stimulatory KIRs in a similar cohort of children with B-ALL (n = 185) and T-ALL (n = 33).⁴  In contrast, they did not find a reduction of stimulatory KIRs in childhood ALL patients, although allele frequencies of stimulatory KIRs between the Canadian and the European study did not differ significantly.⁴ 

Given these observations, we determined KIR genotypes of the 17 known KIR genes of 328 pediatric patients (203 male and 125 female) with ALL (86 B-ALL, 179 common ALL, and 63 T-ALL) treated according to the AIEOP-BFM-ALL 2000 protocol. The majority of patients were of European origin with a frequency being in line with Babor’s work.⁴  The control group consisted of 339 healthy blood donors with a median age of 25 and self-reported Caucasian ethnicity. We used a DNA-based quantitative real-time polymerase chain reaction–based method according to Vilches et al and Alves et al.^5,6 

Analyzing prognostic factors often reveals different and potentially coincidental results in different analyses, often because of the fact that no adjustment for multiple testing was done. Therefore, we carefully accounted for this problem in calculating our sample size and in interpreting our analyses. Power calculation was based on a 2-sample test for proportions using proportions for the 6 KIR genes of main interest given in Almalte et al.³  Adjusting for multiple testing of 6 genes, the level of significance was set to .008 to reach a global significance level of .05 (with a power of .8). Therefore, a minimum of 256 patients and 256 controls had to be included, and the end point of all analyses was the presence of ALL (control vs patient). A possible association between occurrence of KIR genes and age, sex, DNA index, or genetics was not observed (data not shown).

Univariate logistic regression did not reveal any significant difference in frequency distribution of KIR alleles and genotypes in children with ALL compared with healthy controls (Table 1). We also analyzed receptor distributions of the 6 activating receptors among the different leukemic risk categories (low, 35%; intermediate, 46%; and high, 19%). No significant difference in receptor combinations could be found between controls and these 3 subgroups (data not shown). Furthermore, KIR genes were assigned to either KIR haplotype A or B. The distribution of these 2 haplotypes among ALL patients and the healthy reference group was 26.8% vs 30.4% for haplotype A and 73.2% vs 69.6% for haplotype B, respectively. These data are in line with published KIR gene frequencies in a Caucasian population.⁷  We next determined KIR B-content scores for patients and controls and found similar score distributions between the 2 groups. Furthermore, involvement of the central nervous system was independent of activating KIRs, KIR haplotype, and B-content score. An increasing risk for ALL with an increasing number of activating KIR genes was also excluded (P = .173).

Our results are in sharp contrast to the findings of Almalte et al.³  Diversity of KIR gene expression among different ethnic groups has to be taken into account as an explanation for the discrepancy of results. For instance, an increased KIR A/A genotype was found in Hispanic children with ALL, but not in non-Hispanic children.⁸  Moreover, the possibility of coincidental results in statistical analyses might have to be excluded.

In summary, we did not find a significant difference in KIR gene expression between pediatric patients with ALL and healthy controls, and our results do not support the hypothesis that the number of activating KIR genes confers susceptibility or resistance to pediatric ALL. Further research is necessary to elucidate genetic factors involved in the development of ALL in children.