Background

Cure rates in classic Hodgkin lymphoma (cHL) range from 70 to 90 %, but complete recovery of patients is burdened by long-term toxicities of chemoradiation treatment, particularly heart and lung injury and secondary neoplasms [1]. Although in recent years interim [18F]-fluoro-2-deoxy-d-glucose positron emission tomography carried out after 2 cycles of doxorubicin, bleomycin, vinblastine, dacarbazine (ABVD) treatment (iPET-2) has emerged as a powerful predictor of treatment outcome in advanced-stage cHL patients [2], no other biomarkers have been shown capable of predicting treatment outcome at baseline with the same accuracy. Because of its unique histology, cHL is an interesting study model for the identification of new immunological and immunogenetic factors that may confer susceptibility to tumor or influence response to treatment [3]. In fact, the peculiar architecture of “malignant lymphogranuloma” is characterized by the presence of only a few neoplastic cells—Hodgkin-Reed-Sternberg (HRS) cells—growing in a microenvironment rich with immune system cells that are incapable of mounting an effective antitumor response [4].

Despite the considerable amount of evidence suggesting that antitumor immune surveillance is downregulated in cHL [5], few data exist on the impairment of natural killer (NK) cell function in cHL. NK cells are a key component of the innate immune system and play an important role in antitumor surveillance. Their activity is regulated by several receptor families, including killer cell immunoglobulin-like receptors (KIRs) which recognize human leukocyte antigen (HLA) class I molecules expressed on target cells [6]. NK cells express different combinations of the KIR2DL1, KIR2DL2, KIR2DL3, KIR2DL4, KIR2DL5, KIR3DL1, KIR3DL2, and KIR3DL3 inhibitory KIR genes and KIR2DS1, KIR2DS2, KIR2DS3, KIR2DS4, KIR2DS5, and KIR3DS1 activating KIR genes [7]. The KIR gene family has been divided into two broad groups of haplotypes according to gene content. KIR A haplotypes encode the KIR2DL3, KIR2DL1, KIR3DL1, and KIR3DL2 inhibitory genes and the KIR2DS4 activating gene. These haplotypes have identical KIR gene content, but can differ for allelic polymorphism. KIR B haplotypes contain various combinations of both activating and inhibitory KIR genes and show less allelic polymorphism [8, 9].

KIR receptors recognize four mutually exclusive epitopes carried by HLA-A, HLA-B, and HLA-C class I molecules: C1, C2, Bw4, and A3/A11. HLA-C is known to consist of two functional classes of allotypes, depending on an amino acid dimorphism at residue 80 of the α1-helix. The C1 epitope is carried by HLA-C allotypes with asparagine at position 80 and by some HLA-B alleles in the Asian population that have asparagine at position 80 and valine at position 76. The C2 epitope is carried by HLA-C allotypes characterized by a lysine at position 80 of the α1-helix. The Bw4 epitope is carried by HLA-A and HLA-B allotypes that have arginine at position 83. The A3/11 epitope of HLA-A is carried by two HLA-A allotypes, HLA-A*03 and HLA-A*11 [10, 11].

The diversity in gene content of KIR haplotypes and the extensive polymorphism of HLA and KIR genes make it possible for an individual to express KIR genes for which there are no HLA ligands and vice versa, which thus determines extremely variable immunogenetic profiles among individuals [12]. KIR A haplotypes seem to provide better immune surveillance against viral infections and tumor cells, whereas KIR B haplotypes seem to have a favorable role during pregnancy [10].

Although there is a growing interest in the role of NK cells in oncohematologic disorders [13–17], to the best of our knowledge, there is only one previous report investigating the possible involvement of KIR genes in cHL [18]. To gain further insight into the role of NK cells in cHL and their possible impact on response to treatment, we investigated KIR gene frequencies, KIR haplotypes, and KIR-ligand combinations in a cohort of advanced-stage cHL patients compared to healthy controls. Moreover, we evaluated the correlation of KIR genes and KIR haplotypes with the achievement of negative iPET-2 and their possible impact on progression-free survival (PFS).

Methods

Ethics, consent, and permissions

In accordance with the 1975 guidelines of the Declaration of Helsinki and after obtaining approval for the study from the four competent local Ethics Committees (Cagliari, Catania, Cuneo, and Reggio Calabria), the medical staff of each participating center exhaustively informed their patients of the potential risks and benefits of the planned treatment and the need for periodical clinical and laboratory checkups.

Patients, healthy controls, and treatment protocols

One hundred and thirty-five patients with advanced-stage cHL were included in the study. All patients were enrolled in the international prospective multicenter clinical trial HD 0607 (NCT00795613). A complete data set containing all the following parameters was collected at baseline: age, sex, histological subtype, presence of B-symptoms (fever, weight loss exceeding 10 % of body weight in 6 months, night sweats), extranodal involvement or bulky mass, absolute lymphocyte and neutrophil count, hemoglobin and albumin, Ann Arbor stage, and International Prognostic Score (IPS) [19].

All the patients received two courses of standard ABVD treatment before undergoing iPET-2 to assess tumor chemosensitivity. Patients were stratified into two groups according to the results of iPET-2. Patients with positive iPET-2 were randomly assigned to receive four courses of escalated BEACOPP (rituximab, bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, prednisone) or R-BEACOPP (BEACOPP supplemented with rituximab), followed by four courses of baseline BEACOPP or R-BEACOPP. Patients with a negative iPET-2 scan continued with four more courses of ABVD. Patients in complete remission with a negative final PET scan who presented bulky disease at baseline or showed a PET-negative residual mass at the end of chemotherapy were randomly assigned to receive consolidation radiotherapy or no further treatment.

A group of 221 healthy subjects (104 males and 117 females) with a mean age of 35 (range 25–55 years) were selected for control purposes.

HLA and KIR typing

HLA typing was performed in healthy controls and patients using commercially available PCR kits with sequence-specific primers (HLA-A-B-C SSP Combi Tray, Olerup, Stockholm, Sweden, and Micro SSP™ Generic HLA class I typing tray SSP1L, One Lambda, Inc., Canoga Park, CA, USA). HLA-C groups C1 and C2 and HLA-Bw4 were determined by high-resolution typing for the alleles at the HLA-A, HLA-B, and HLA-C loci using a polymerase chain reaction-sequence-specific primer (PCR-SSP) method according to the manufacturers’ instructions (allele-specific PCR-SSP kits: Olerup SSP AB, Stockholm, Sweden).

Distribution of KIR genes (KIR2DL1, KIR2DL2, KIR2DL3, KIR2DL4, KIR2DL5A, KIR2DL5B, KIR2DS1, KIR2DS2, KIR2DS3, KIR2DS4, KIR2DS5, KIR3DL1, KIR3DL2, KIR3DL3, KIR3DS1) were determined by PCR-SSP using a previously reported method [20, 21].

Patients were stratified into two groups according to homozygosity for KIR A haplotype (KIR genotype AA) and heterozygosity or homozygosity for KIR B haplotype (KIR genotypes AB and BB, referred together as KIR genotype Bx) [21, 22].

Statistical analysis

KIR gene frequencies, KIR haplotypes, and KIR-ligand combinations were compared between cHL patients and healthy controls as well as between cHL patients who achieved negative iPET-2 and those who did not. The abovementioned set of independent variables was included in univariate and multivariate analyses. Where appropriate, significant differences were calculated using Fisher’s two-sided exact test or Pearson’s chi-squared test. Variables with a p value lower than 0.2 in univariate analysis were included in multivariate analysis using a multi-step forward binary logistic regression model, where negative iPET-2 scan was considered a dependent variable. Only p values ≤0.05 were considered to be statistically significant. The probability of achieving overall survival (OS) and PFS was calculated using the Kaplan-Meier method. Comparisons were also made to assess PFS in patient subgroups according to KIR haplotype or iPET results. The log-rank test was used to compare the groups of patients.

Results

Characteristics of patients and controls

The demographic and clinical characteristics of 135 advanced-stage cHL patients (67 males and 68 females, mean follow-up 35 months, range 8–60) and 221 controls are shown in Table 1. The overall mean age at diagnosis was 34 years (range 15–69). The large majority of patients (80 %) had a nodular sclerosing histological subtype. Bulky mass and “B” symptoms were found in 40.7 and 54.1 %, respectively. According to the HD 0607 inclusion criteria, all patients belonged to advanced Ann Arbor stages (IIb–IV) with an unfavorable IPS in 28.9 % of cases. The demographic characteristics of the 221 healthy controls (104 males and 117 females) with a mean age of 35 (range 25–55 years) were not statistically different from the cHL cohort.

Table 1 Characteristics of 135 patients with advanced-stage classic Hodgkin lymphoma and 221 healthy controls
Full size table

KIR gene frequencies, KIR haplotypes, and KIR-HLA ligand combinations in cHL patients and controls

Table 2 shows KIR gene frequencies, KIR haplotypes, and KIR-HLA ligand combinations in 135 cHL patients and 221 healthy controls. No significant differences were found between patients and controls for the frequencies of the single activating and inhibitory KIR genes. Overall, the healthy controls had significantly higher frequencies of the HLA-C1 ligand (C1/C1 or C1/C2) compared to patients (72.7 vs. 60 %, p = 0.014); on the other hand, frequency of the HLA-C2 ligand (C2/C2 or C1/C2) was significantly higher in cHL patients (88.9 vs. 76 %, p = 0.003). No significant differences were found for frequencies of the HLA-Bw4 ligand within the two groups. Homozygosity for KIR A haplotype combined with homozygosity for the HLA-C1 ligand (KIRAA/C1C1) was significantly higher in healthy controls (15.7 vs. 4.8 %, p = 0.001). The analysis of the combinations of KIR genes and their specific ligands showed a significantly higher frequency of KIR3DS1 and KIR3DL1 when combined with the absence of their specific ligand HLA-Bw4 in cHL patients (15.5 vs. 6.3 %, p = 0.006 and 33.3 vs. 21.3 %, p = 0.013, respectively).

Table 2 KIR and HLA gene frequencies, KIR haplotypes, and KIR-HLA ligand combinations in 135 advanced-stage classic HL patients and 221 healthy controls
Full size table

KIR gene frequencies, KIR haplotypes, and KIR-HLA ligand combinations in cHL patients with positive or negative iPET-2 results

Table 3 shows HLA and KIR gene frequencies, KIR haplotypes, and the combinations between KIRs and their ligands in 33 patients who showed positive iPET-2 in comparison with 102 patients with negative iPET-2.

Table 3 KIR and HLA gene frequencies, KIR haplotypes, and KIR-HLA ligand combinations in 135 advanced-stage classic Hodgkin lymphoma patients according to positive or negative interim PET-2 results
Full size table

A significantly higher frequency of the KIR2DS2 activating KIR gene was found in patients with positive iPET-2 (81.8 vs. 50 %, p = 0.001). Homozygosity for KIR A haplotype was present with a significantly higher frequency in patients with negative iPET-2 (36.3 vs. 15.2 %; p = 0.03). In particular, only 5 of the 33 patients with positive iPET-2 had KIR genotype AA, whereas the remaining 28 (84.8 %) had KIR genotype Bx. Homozygosity for KIR A haplotype remained the only predictive factor for the achievement of negative iPET-2 (p = 0.039). The analysis of HLA-KIR combinations showed that cHL patients with a positive iPET-2 scan had higher frequencies of the activating KIR genes, KIR2DS1 and KIR2DS2, combined with the presence of their specific ligands C2 and C1, respectively (KIR2DS1/HLA-C2: 66.7 vs. 35.3 %, p = 0.03; KIR2DS2/HLA-C1: 63.3 vs. 20.6 %, p = 0.004). Higher frequencies were also observed in this group of patients for the inhibitory KIR genes, KIR2DL1 and KIR2DL2, combined with the presence of their specific ligands C2 and C1, respectively (KIR2DL1/HLA-C2: 81.8 vs. 57.8 %, p = 0.021; KIR2DL2/HLA-C1: 62.7 vs. 32.3 %, p = 0.018).

Overall survival and progression-free survival following treatment and according to KIR haplotype and iPET-2

In the cohort of 135 patients, 5-year OS and PFS was 93.6 and 79 %, respectively (Fig. 1). PFS was significantly higher in patients who achieved negative iPET compared to those who did not (89.8 vs. 55.8 %, p = 0.0001); among patients with negative iPET, those homozygous for KIR A haplotype had higher PFS rates compared to those carrying Bx haplotypes, although the difference was not statistically significant (92.3 vs. 85.3 %) (Fig. 2).

Fig. 1
figure 1

Overall survival (OS) and progression-free survival (PFS) in the cohort of 135 patients. CI confidence interval

Full size image
Fig. 2
figure 2

Progression-free survival (PFS) in the cohort of 102 patients with negative interim PET, stratified in two groups according to KIR haplotype (p = NS). CI confidence interval

Full size image

Discussion

Survival for advanced-stage cHL patients has increased to 70–85 % over the past years, but the optimal balance of the risks and benefits of treatment is still a matter of debate. At 15 years from the end of treatment, the risk of dying from lymphoma is less than the risk of dying from therapy-related complications [23]. The main dilemma remains whether frontline intensive chemotherapy should be given to all patients or be restricted to patients who are poor responders to less intensive treatment approaches. So far, iPET-2 scan represents the only strong predictor of response to treatment [2]. There are currently no other available prognostic biomarkers capable of distinguishing patients requiring more or less intensive treatment. For this reason, many ongoing studies focus their efforts on the definition of new predictive factors for prognosis and response to treatment in cHL. Ideally, risk assessment should combine innovative predictive biomarkers with new imaging technologies [24–27].

Classic HL represents an extremely interesting study model for the assessment of immunologic and immunogenetic factors that may confer susceptibility to tumors or facilitate tumor immune escape mechanisms [4, 5]. Currently few data exist on the role of impaired NK cell function in immune escape of HRS cells.

Besson et al. investigated association between KIRs and cHL in a family study [18]. They provide evidence that activating KIR genes, in particular KIR2DS1 and KIR3DS1, may exert protection in cHL. In our cohort of patients, we found a significantly higher frequency for KIR3DS1 in the absence of its specific Bw4 ligand, suggesting that this activating KIR gene cannot properly exert its protective function. No other significant differences were observed between cHL patients and controls for single KIR gene frequencies. Vice versa, when we analyzed C1 and C2 epitopes, we found that C2 had a significantly higher frequency in cHL patients (88.9 vs. 76 %, p = 0.008). Conversely, the C1 epitope was significantly higher in healthy controls (72.7 vs. 60, p = 0.036) (Table 2). Interestingly, homozygosity for KIR A haplotype and HLA-C1 ligand (KIRAA/C1C1) was significantly higher in healthy controls, suggesting a possible protective role in cHL (15.7 vs. 4.8 %, p = 0.001) (Table 2).

In a recent review, Parham and Moffett suggest that homozygosity for HLA-C1 and KIR haplotype A could be associated to a better immune response with resistance to viral infection [10]. It is possible that the fixed gene content of KIR A haplotypes with their highly polymorphic receptors provides better immune surveillance against viral infections, whereas KIR B haplotypes with their greater variation in gene content and moderate allelic polymorphism could have an important role in immune-tolerance during pregnancy [10]. Our data seem to suggest a protective role for the KIRAA/C1C1 genotype in cHL.

In our study, KIR A haplotype was the only factor significantly associated to a negative iPET-2 result suggesting its favorable role in the achievement of better response to standard chemotherapy. Moreover, the combination of negative iPET-2 with homozygosity for KIR A haplotype seems to represent a favorable combination for identifying patients responsive to standard courses of chemotherapy (PFS = 92.3 %) (Fig. 2).

Finally, we found that cHL patients with iPET-2 positivity had higher frequencies of the KIR-ligand combinations KIR2DS1/HLA-C2 and KIR3DS2/HLA-C1. The interaction between KIR2DS1, KIR2DS2, and their HLA-C ligands might increase the production of tumor growth factor (TGF)-β by NK cells. It has been suggested that TGF-β may reduce tumor lysis mediated by NK cells and stimulate the production of proinflammatory cytokines such as tumor necrosis factor alpha (TNF-α) and interferon gamma (IFN-γ) [28].

Conclusions

Our results, though preliminary, seem to show that homozygosity for KIR A haplotype, alone or in combination with homozygosity for the HLA-C1 KIR ligand, confers protection to cHL. The association found for the KIR-AA genotype and achievement of negative iPET-2 suggests that testing for KIR haplotypes could be used in clinical practice to enhance the power of iPET-2 in predicting better treatment outcome in cHL patients. Overall, our findings may represent a rationale for developing new risk-adapted treatment strategies based on the combination of KIR biomarkers and PET scans. They furthermore strengthen the hypothesis of innovative and targeted therapies aimed at modulating NK cell function. The use of anti-inhibitory-KIR antibodies could prove to be an effective way of blocking the binding interface of inhibitory KIRs with HLA class I ligands expressed by lymphoma cells, thus preventing a tolerogenic interaction and augmenting NK cell cytotoxicity [28–31].