Polymorphisms in the genes ERCC2, XRCC3 and CD3EAP influence treatment outcome in multiple myeloma patients undergoing autologous bone marrow transplantation

Abstract

Individual variations in the ability to cope with DNA damage by DNA repair may be essential for the response to chemotherapy, since cancer cells from patients with an effective DNA repair may survive treatment. We have studied the effect on time to treatment failure (TTF) and overall survival (OS) of polymorphism in the DNA repair genes ERCC1, ERCC2 and XRCC3, and in the apoptotic genes PPP1R13L and CD3EAP in 348 patients with multiple myeloma undergoing autologous bone marrow transplantation. Carriers of the variant C-allele of ERCC2 K751Q, the variant T-allele of XRCC3 T241M and the variant A-allele of CD3EAP G-21A had a 1.3-fold, 1.8-fold and 1.9-fold longer TTF, respectively, than homozygous wild type carriers (p = 0.006, p = 0.004, p < 0.001). The polymorphism CD3EAP G-21A also had significant effect on OS (p < 0.045). The polymorphism ERCC2 K751Q may to be related to sex, since the prolonged TTF was only seen in women (p = 0.001). Carriers of the combination of variant alleles of ERCC2 K751Q and XRCC3 T241M had 2.8-fold longer TTF (p = 0.0002). This indicates that suboptimal repair of both DNA mechanisms favors prolonged TTF and that polymorphism in ERCC2, XRCC3 and CD3EAP predicts the outcome for patients treated with autologous stem cell transplantation. © 2006 Wiley-Liss, Inc.

It is becoming increasingly clear that interindividual variation in the response to chemotherapy in cancer is important. In this report we have attempted to elucidate whether polymorphism in DNA repair genes influences the outcome in patients with multiple myeloma (MM).

Younger patients with MM are usually treated with high-dose therapy with melphalan, followed by autologous stem cell transplantation (ASCT), which has resulted in significantly increased overall survival (OS) and quality of life.1 However, time to treatment failure (TTF) varies considerably among patients. Because DNA repair is essential for correcting damages caused by DNA damaging therapy, individual variation in DNA repair capacity may be central in resistance to chemotherapy. Cells from MM patients who relapse after treatment with melphalan exhibit increased repair capacity when compared to cells from patients without previous treatment.2 Moreover, genetic variations that modify DNA repair capacity influence the outcome of chemotherapy.3, 4, 5

The genes ERCC1 and ERCC2 are involved in the nucleotide excision repair (NER), which is responsible for repair of bulky DNA damages caused by chemotherapy.6, 7ERCC1 encodes a subunit of the endonuclease, which makes the incision 5′ of the DNA damage in NER.6ERCC2 encodes a subunit of the transcription factor TFIIH, which is essential for both transcription and NER.6 The polymorphisms ERCC2 N312D and ERCC2 K751Q result in amino acid substitutions (Fig. 1) and the variant alleles have been associated with increased levels of DNA adducts and low DNA repair capacity.8, 9XRCC3 (located on chromosome 14q32) is required for efficient repair of double strand breaks via homologous recombination repair, correct chromosomal segregation, as well as the repair of DNA cross links.10PPP1R13L encodes an inhibitor of the RelA subunit of the transcription factor NF-κB, which has a central function in apoptotic regulation in most cancers as well as in MM.11 Changes in the PPP1R13L level have been shown to modify cisplatin-induced apoptosis.12 The 3′-end of CD3EAP, also named CAST and ASE-1, (antisense to ERCC1) is overlapping with the ERCC1 transcript.13CD3EAP is an ubiquitous nuclear protein, which may be involved in T-cell activation.14 Homozygous carriers of the haplotype ERCC1 N118N^A, CD3EAP G-21A^G and PPP1R13L IVS1 A4364G^A have been shown to be at a much higher risk for basal cell carcinoma,15, 16 breast cancer and lung cancer.17, 18 The haplotype was not associated with risk of testis cancer or lung cancer in males.18, 19 We hypothesized that polymorphisms, which may modify DNA repair capacity and apoptosis, would modify the response to chemotherapy. In a population of 348 patients with MM treated with high dose chemotherapy we have studied the effect on TTF and OS of the DNA repair gene polymorphisms ERCC2 K751Q, ERCC2 D312N, XRCC3 T241M and ERCC1 N118N, and 2 other polymorphisms, PPP1R13L IVS1 A4364G and CD3EAP A-21G, which together with ERCC1 N118N, define the previously studied haplotype. We report that 2 DNA repair gene polymorphisms, ERCC2 K751Q and XRCC3 T241M, and the polymorphism CD3EAP G-21A significantly influence the outcome of high-dose chemotherapy; for CD3EAP G-21A and ERCC2 K751Q the effect may be contained in women only.

Figure 1. Graphical presentation of the studied SNPs on chromosome 19. The ERCC1 polymorphism is silent and does not result in an amino acid substitution. The CD3EAP polymorphism is G to A transversion in the 5′ untranslated region of the mRNA. The PPP1R13L polymorphism is in intron 1 of the PPP1R13L gene. The ERCC2 D312N polymorphism gives rise to the Asp to Asn amino acid substitution in position 312 of ERCC2. ERCC2 K751Q gives rise to the Lys to Gln substitution in position 751 in ERCC2.

ASCT, autologous stem cell transplantation; CD3EAP, CD3e molecule, epsilon associated protein (also named ASE-1, antisense to ERCC1); CI, confidence interval; ERCC1, excision-repair cross-complementing 1; ERCC2, excision-repair cross-complementing 2; MM, multiple myeloma; NER, nucleotide excision repair; OS, overall survival; PPP1R13L, protein phosphatase 1, regulatory (inhibitor) subunit 13 like (also named RAI, RelA associated inhibitor); RR, relative risk; TTF, time to treatment failure; XRCC3, X-ray repair cross-complementing group 3.

Material and methods

Subjects

Patients were recruited from August 1994 to August 2004 from 4 participating Danish centers in Denmark with a catchment population of about 5 million inhabitants. One hundred eighty-five patients diagnosed with MM in high-dose treatment protocols including ASCT (NMSG no. 5/94, 7/98 and 11/00) administrated by the Nordic Myeloma Study Group (NMSG),20 and 163 patients treated identically, but not registered in the NMSG protocols, were included.

The study was approved by the Danish Ethical Committee (01-158/03).

Clinical data and response criteria

The diagnosis of MM was accepted if criteria A + C, A + D or B + C + D of the following were attained. Three patients with smoldering myeloma were included (C +D). A: serum monoclonal compartment concentration of immunoglobulin IgG > 30 g/L, IgA > 20 g/L, the presence of an M-protein of IgD or IgE regardless of concentration, or Bence-Jones proteinuria > 1 g/24 hr; B: M-protein in serum or urine at a lower concentration than described under A; C: at least 10% plasma cells in bone marrow aspirate or biopsy-verified plasmacytoma of bone or soft tissue; and D: osteolytic bone lesions. Staging was according to Durie and Salmon.21

Progression was defined by a more than 25% increase in serum M-protein or 25% increase in immunoglobulin levels above upper normal levels, confirmed by 2 separate measurements within a 1-month interval. Increase of Bence-Jones proteinuria to more than 1.0 g/24 hr or other signs of progression such as hypercalcemia, progressive skeletal disease or soft-tissue plasmacytoma were also considered as progression. In patients for whom progression could not be evaluated from the above-mentioned criteria, progression was defined as a 25% increase in infiltration of plasma cells in bone marrow. Occurrence of secondary malignancies and death without progression was considered as events not related to progression. TTF and OS were calculated from date of transplantation to date of progression or death. Eleven patients died during transplantation. Two patients died from other cancer forms than from MM, and 2 patients died of other causes. These patients were included in the analysis of OS, but they were censored at time of death in the analysis of TTF. Sixty-eight patients, including the 11 patients who died during the transplantation procedure, were followed up less than 2 years.

Eligibility criteria

NMSG no. 5/94: All patients younger than 60 years could be included, provided they were not considered ineligible for induction therapy because of severe chronic heart or lung disease, other active malignancy or severe coincident illness, psychiatric disease or abuse, terminal illness or refusal. NMSG no. 7/98 and NMSG no. 11/00: As described in NMSG no. 5/94, but with inclusion of all patients younger than 65 years. Patients not included in the above-mentioned high-dose treatment protocols were submitted to ASCT in accordance to the judgment of eligibility by the treating physician.

NMSG treatment protocols

NMSG no. 5/94: Induction therapy with VAD (vincristine, doxorubicin and dexamethasone; peripheral blood stem cell harvest at regeneration after cyclophosphamide, high-dose chemotherapy with melphalan followed by ASCT, maintenance therapy with interferon. Patients were transplanted only if they had an initial response to chemotherapy before harvesting.1 NMSG no. 7/98: As described for NMSG no. 5/94, but patients were included in the study regardless of their response to the induction treatment.22 NMSG no. 11/00: The treatment strategy was the same as for NMSG no. 7/98 except for the induction treatment. The patients were randomized to treatment with either 3 series of VAD or 2–3 series of cyclophosphamide (1 g/m²) combined with dexamethasone (40 mg daily for 4 days). Twenty-nine patients were treated with VAD and 33 patients were treated with cyclophosphamide. There was no difference in TTF and OS in the 2 treatment groups.

The patients not included in the above-mentioned high-dose treatment protocols were treated according to the judgment by the treating physician. Standard induction chemotherapy for this group of patients was VAD or cyclophosphamide (1 g/m²) and dexamethasone (40 mg daily for 4 days). Induction treatment was followed by peripheral blood stem cell harvest at regeneration after cyclophosphamide, high-dose chemotherapy with melphalan followed by ASCT, maintenance therapy with interferon.

Human tissue samples

Peripheral blood mononuclear cell was purified from 292 leukapheresis products by buffy coat preparation. From 56 patients 10 μm sections were collected 10 times from paraffin-embedded bone marrow samples. Material was not available for 19 patients undergoing autologous bone marrow transplantation and these patients were not included in the study.

DNA purification

DNA for analysis was purified from PMBC by the salting out method23 or from paraffin-embedded tissue by phenol extraction as described in Ref.24.

Detection of single nucleotide polymorphisms

The polymorphisms are presented in Figure 1. The dbSNP web address is www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=snp

Genotypes were determined on an ABI 7500 using endpoint readings. Ten microliter reactions contained about 50 ng DNA, 5 μL mastermix (Applied Biosystems, Birkerød, Denmark), 100 nM of each probe and 900 nM primers. Controls were included in each run, and 10–50% of the samples was retyped for reproducibility and gave 99–100% identical results. Moreover, for 10 persons, both DNA from bone marrow and from leukapheresis products were typed with identical results. ERCC2 K751Q (rs1052559) (XPD exon23, A→C) and ERCC2 D312N (rs1799793) (XPD exon10, G→A) were determined as previously described.25PPP1R13L VS1 A4364G (rs1970764) (RAI intron1, A→G) was determined as previously described in Ref.19. CD3EAP G-21A (rs967591) (ASE-1, exon1 G→A) was determined using the primers 5′-TCTGCAACCTGGTGCGAG-3′, 5′-CCTTTCTCCTTCCACCAACG-3′ and the probes G-allele: 5′-FAM-AGGGTTGCCTGAGGTGTGGGTCC-BHQ, A-allele: Yakima Yellow-AGG GTTACCTGAGGTGTGGGTCC-BHQ-3′. The reactions were run for 40 cycles for 15 sec at 94°C, 60 sec at 62°C. ERCC1 N118N (rs3177700) (ERCC1 exon4, A→G) was determined using the primers 5′-GATGGCTTCTGCCCTTCG T-3′ and 5′-GGGAATTACGTCGCCAAATTC-3′ and the probes G-allele: 5-Fam-CGTGCGCAACGTGCCC-BHQ-3′, A-allele: 5′-Yakima Yellow-CGTGCGCAATGTGCCCTG-BHQ-3′ (TAGCopenhagen, Copenhagen, Denmark). The reactions were run for 40 cycles for 15 sec at 94°C, 60 sec at 63°C. XRCC3 T241 M (rs861535) (XRCC3 18067, C→T) was determined as previously described.26

A varying number of PCR reactions failed for each genotype. These PCR reactions were repeated using 1/10 the amount of DNA and 3 times the amount of DNA. If 3 independent attempts failed, the genotype was left undetermined. Nearly 40% of the PCR reactions for XRCC3 T241M failed, probably because the assay is very sensitive to iron-containing impurities of hemoglobin from the DNA purification step.

Statistical methods

SPSS statistical software was used for all calculations (SPSS for Windows, Rel. 14.0.0. 2005, SPSS, Chicago). All tests were 2-sided and p values < 0.05 were regarded as significant.

Fisher's exact test was used for comparing categorical variables and Mann–Whitney test was used for comparing continuous with categorical variables. Kaplan–Meier method and the log rank test were used to compare TTF and OS between groups. The Cox proportional hazards model, log-likelihood statistics, was applied for univariate analyses of covariates and for multivariate analysis. Significant variables with a p value < 0.05 by univariate analysis were included in the multivariate Cox analyses to identify variables of independent significance. In accordance to the Bonferroni correction for multiple comparisons the p value was lowered from 0.05 to 0.004.

Results

Study population

Table I lists the baseline demographic and clinical characteristics of the patients included in the study. In the present study group, age and albumin were not significantly associated with TTF and survival, probably because only patients eligible to high-dose treatment were included (Table II). Advanced stage according to Durie–Salmon as well as ISS27 and elevated levels of creatinine were associated with poor OS (Table II). β2-Microglobulin was a prognostic marker for poor TTF and OS. Progression of disease occurred in 58.3% of the cases and 42.0% of the patients died during the study period. The study included patients censored according to the high-dose treatment protocols of NMSG, as well as uncensored patients treated with high-dose chemotherapy followed by ASCT. Statistical analysis showed no difference in TTF and OS between the 3 different treatment protocols, between patients treated with VAD or cyclophosphamide and dexamethazone and the uncensored patients not included in the protocols.

Table I. Patient Characteristics at the Time of Diagnosis of 348 Patients

	No. of patients	Median	Range
1 Vaues in parentheses are percentages. 2 Months from bone marrow transplantation. 3 Number of events. 4 Censored pt.: exclusion of patients with occurrence of secondary malignancies and death without progression.
Sex
Male	204 (59)1
Female	144 (41)
Age		56	28–69
<60	246 (71)
≥60	101 (29)
β2-Microglobulin (mg/dL)	249	3.9	1.2–57
Creatinine (μmol/L)	332	98	47–833
Albumin (g/dL)	297	3.5	0.25–5.3
ISS
I	56 (23)
II	94 (39)
III	89 (37)
Durie-Salmon stage
I	35 (10)
II	77 (23)
III	228 (67)
TTF (mo)2	201 (58)3	27.4	2.2–117
OS (mo)2	146 (42)3	63.0	0.5–128
TTF follow-up time (mo)4 [censored pt. only]		29.3	2.3–116.6
TTF follow-up time (mo)4 [all]		19.6	2.2–116.6
OS follow-up time (mo)4 [censored pt. only]		45.5	8.6–128.2
OS follow-up time (mo)4 [all]		35.0	0.4–128.2

Table II. Univariate Analysis of the Effect of Known Prognostic Markers on TTF and OS

	TTF		OS
	HR	p value	HR	p value
1 β2-Microglobulin and creatinine were treated as continuous variables and log transformed. The hazard ratio indicates risk when values are doubled. 2 Values in parentheses are 95% CI. 3 Albumin was treated as a continuous variable. The hazard ratio indicates risk when value increased 1 g/dL.
β2-Microgloblin1	1.3 (1.1–1.5)2	0.007	1.5 (1.3–1.8)	<0.0001
Albumin3	0.9 (0.7–1.1)	0.25	0.9 (0.7–1.1)	0.25
Creatinine2	1.1 (1.0–1.3)	0.17	1.5 (1.2–1.8)	<0.0001
ISS		0.09		0.002
(I vs.II)	1.3 (0.8–2.0)		1.3 (0.7–2.3)
(I vs. III)	1.7 (1.0–2.7)		2.4 (1.4–4.2)
Age (<60 vs. ≥60)	0.9 (0.6–1.2)	0.52	1.4 (1.0–2.0)	0.09
Sex (female vs. male)	1.2 (0.9–1.6)	0.17	1.1 (0.8–1.6)	0.49
Durie–Salmon		0.36		0.026
(I vs. II)	1.3 (0.7–2.2)		2.5 (1.1–5.3)
(I vs. III)	1.4 (0.9–2.3)		2.3 (1.1–4.8)

Single nucleotide polymorphism

Genotypes of ERCC2 K751Q, ERCC2 D312N, PPP1R13L IVS A4364G, CD3EAP G-21A, ERCC1 N118N and XRCC3 T241M were determined. There was no difference in the allele frequencies among patients dependent on the participating centers, and the allele frequencies of the patients with MM were similar to those previously observed for controls in the Danish Diet, Cancer and Health cohort,17, 18, 25, 26 indicating that the polymorphisms in this subgroup of patients were not associated with risk of MM (results not shown). The 5 polymorphisms which are located on chromosome 19q13.2-3 were in linkage disequilibrium. The linkage patterns were exactly the same as previously reported for Danish study populations.17

In univariate analyses the effect of genotype on TTF and OS was determined (Table III). Data adjusted for all other survival-related factors (β2-microglobulin, creatinine and Durie–Salmon) are shown in parentheses and gave similar results.

Table III. Univariate Analysis of the Effect of Genotypes on TTF and OS

Genotype	N	%	Median TTF (mo)	p	HR	Median OS (mo)	p	HR
1 Values in parentheses are 95% CI. 2 Values in italic are adjusted for prognostic related factors (TTF: β2-microglobulin. OS: β2-microglobulin, creatinine and Durie-Salmon). 3 Median survival not reached.
ERCC2 K751Q
AA	138	41	23.8			59.3
AC	135	40	31.5	0.007 (0.007)2	0.7 (0.5–0.9)1 (0.6 (0.4–0.9))s	65.7	0.26 (0.14)	0.8 (0.6–1.2) (0.7 (0.4–1.1))
CC	63	19	31.6	0.07 (0.018)	0.7 (0.5–1.0) (0.5 (0.3–0.9))	94.0	0.24 (0.21)	0.8 (0.5–1.2) (0.7 (0.4–1.2))
AC + CC	198	59	31.5	0.004 (0.003)	0.7 (0.5–0.9) (0.6 (0.4–0.8))	65.9	0.17 (0.11)	0.8 (0.6–1.1) (0.7 (0.5–1.1))
ERCC2 D312N
GG	123	40	23.9			65.9
AG	148	48	28.6	0.11 (0.13)	0.8 (0.6–1.1) (0.7 (0.5–1.1))	63.1	0.69 (0.75)	1.1 (0.7–1.6) (1.1 (0.7–1.7))
AA	40	13	31.6	0.12 (0.11)	0.7 (0.5–1.1) (0.6 (0.3–1.1))	94.0	0.20 (0.69)	0.7 (0.4–1.3) (0.9 (0.4–1.8))
AA + GG	188	60	30.9	0.059 (0.073)	0.8 (0.6–1.0) (0.7 (0.5–1.0))	65.7	0.91 (0.89)	1.0 (0.7–1.4) (1.0 (0.7–1.6))
PPP1R13L IVS1 A4364G
AA	214	65	26.4			55.4
AG	108	33	30.3	0.51(0.61)	0.9 (0.7–1.2) (0.9 (0.6–1.3))	83.9	0.19 (0.11)	0.7 (0.5–1.1) (0.7 (0.4–1.1))
GG	7	2	45.4	0.54 (0.82)	0.7 (0.2–2.7) (0.8 (0.2–3.5))	–3	0.40 (0.46)	0.4 (0.1–3.1) (0.5 (0.1–3.6))
AG + GG
	115	35	30.3	0.46 (0.59)	0.9 (0.7–1.2) (0.9 (0.6–1.3))	83.8	0.07 (0.09)	0.7 (0.5–1.0) (0.7 (0.4–1.1))
CD3EAP G-21A
GG	225	67	24.6			57.7
AG	105	31	45.9	0.001 (0.002)	0.6 (0.4–0.8) (0.5 (0.4–0.8))	91.1	0.045 (0.06)	0.7 (0.5–1.0) (0.7 (0.4–1.1))
AA	7	2	13.5	0.61 (0.85)	1.3 (0.5–3.6) (1.1 (0.4–3.5))	28.7	0.75 (0.34)	0.8 (0.2–3.3) (0.4 (0.1–2.8))
AG + AA	112	33	45.9	0.002 (0.002)	0.6 (0.4–0.8) (0.6 (0.4–0.8))	–3	0.045 (0.045)	0.7 (0.5–1.0) (0.6 (0.4–1.0))
ERCC1 N118N
AA	130	40	26.0			59.5
AG	141	44	30.8	0.46 (0.54)	0.9 (0.6–1.2) (0.9 (0.6–1.3))	74.7	0.22 (0.60)	0.8 (0.5–1.1) (0.9 (0.6–1.4)
GG	51	16	32.9	0.16 (0.05)	0.7 (0.5–1.1) (0.5 (0.3–1.0))	65.7	0.33 (0.13)	0.3 (0.5–1.3) (0.6 (0.3–1.2))
AG + GG	192	60	30.9	0.24 (0.19)	0.8 (0.6–1.1) (0.8 (0.5–1.1))	74.7	0.16 (0.28)	0.8 (0.6–1.1) (0.8 (0.5–1.2))
XRCC3 T241M
CC	110	45	23.9			65.7
CT	102	41	45.4	0.004 (0.015)	0.6 (0.4–0.9) (0.6 (0.3–0.9))	–3	0.27 (0.23)	0.8 (0.5–1.2) (0.7 (0.4–1.3))
TT	31	13	44.6	0.14 (0.15)	0.7 (0.4–1.1) (0.6 (0.2–1.2))	81.7	0.71 (0.26)	0.9 (0.5–1.7) (0.6 (0.2–1.5))
CT + TT	133	55	45.4	0.003 (0.01)	0.6 (0.4–0.8) (0.6 (0.4–0.9))	81.6	0.29 (0.15)	0.8 (0.5–1.2) (0.7 (0.4–1.2))

Carriers of the variant C-allele of ERCC2 K751Q had a median TTF of 31.5 month when compared to 23.8 month TTF of homozygous carriers of the wildtype A-allele (p = 0.004) (Fig. 2a). There was no effect on OS. Carriers of the variant A-allele of CD3EAP G-21A had a median TTF of 45.9 month compared with 24.6 month of homozygous wild type G-allele carriers (p = 0.001). Carriers of the variant allele carriers also lived longer than homozygous wild type allele carriers (99.1 month compared to 57.7 month, p = 0.045) (Fig. 2b). Carriers of the variant T-allele of XRCC3 T241M had a median TTF of 45.4 month compared to 23.9 month TTF of homozygous carriers of the wild type A-allele (p = 0.003) (Fig. 2c). No statistically significant effects of ERCC2 D312N, PPP1R13L IVS A4364G and ERCC1 N118N genotypes on TTF or OS were observed. The previously defined high-risk haplotype (ERCC1 N118N^A, CD3EAP G-21AG, PPP1R13L IVS1 A4364GA) was not associated with either differences in TTF or in OS.

Figure 2. Effect on TTF of polymorphisms (a) ERCC2 K751Q, (b) CD3EAP G-21A and (c) XRCC3 T241M. Kaplan–Meier plots of TTF. The numbers at risk at 0, 24, 48 and 72 months are presented below the figure. Wildtype carriers (red line) and carriers of the variant haplotype (blue line).

The effect of sex and genotypes on TTF and survival was investigated (Table IV). For the polymorphisms CD3EAP G-21A and ERCC2 K751Q, significantly prolonged TTF was only observed among women (p = 0.006 and 0.001). A tendency towards longer TTF was seen among men who were carriers of the variant A-allele of CD3EAP G-21A. ERCC2 K751Q was associated with significant prolonged OS among women (p = 0.0005) as well.

Table IV. Univariate Subtype Analysis of the Effect of Polymorphisms CD3EAP G-21 A and ERCC2 K751 Q on TTF and OS in Men and Women

Genotype	Median TTF (mo)	p	HR	Median OS (mo)	p	HR
1 Values in parentheses are 95% CI. 2 Values in italic are adjusted for prognostic-related factors (TTF: β2-microglobulin. OS: β2-microglobulin, creatinine and Durie-Salmon). 3 Median survival not reached.
CD3EAP G-21A
All
GG	24.6			57.7
AG + AA	45.9	0.002 (0.002)2	0.6 (0.4–0.8)1 (0.5 (0.4–0.8))	–3	0.045 (0.045)	0.7 (0.5–1.0) (0.6 (0.4–1.0))
Men
GG	25.1			55.4
AG + AA	31.6	0.13 (0.01)	0.7 (0.5–1.1) (0.5 (0.3–0.9))	–3	0.24 (0.03)	0.7 (0.5–1.2) (0.5 (0.3–1.0))
Women
GG	23.5			65.6
AG + AA	48.6	0.006 (0.041)	0.5 (0.3–0.8) (0.5 (0.3–1.0)	76.5	0.071 (0.3)	0.6 (0.3–1.1) (0.7 (0.3–1.4))
ERCC2 K751Q
All
AA	23.8			59.3
AC + CC	31.5	0.004 (0.003)	0.7 (0.5–0.9) (0.6 (0.4–0.8))	65.9	0.17 (0.11)	0.8 (0.6–1.1) (0.7 (0.5–1.1))
Men
AA	25.1			81.6
AC + CC	28.6	0.30 (0.33)	0.8 (0.6–1.2) (0.8 (0.5–1.3))	53.1	0.37 (0.67)	1.2 (0.8–1.9) (1.1 (0.6–2.0))
Women
AA	16.0			40.1
AC + CC	46.8	0.001 (0.0004)	0.5 (0.3–0.8) (0.3 (0.2–0.6))	74.6	0.0005 (0.001)	0.4 (0.2–0.7) (0.001)

Combination analyses of single nucleotide polymorphisms

Alkylating cytostatics, such as cyclophosphamide and melphalan, used in the high-dose treatment protocol for MM, result in multiple DNA damages, including bulky adducts and crosslinks. It is unclear to what extent the different DNA repair mechanisms are involved in repair of these damages. The 3 polymorphisms (ERCC2 K751Q, CD3EAP G-21A and XRCC3 T241M) that were associated with differences in TTF were combined, and the effect on TTF of combinations of 2 genotypes was investigated in a univariate analysis (Table V and Fig. 3).

Figure 3. Kaplan–Meier plots of TTF using combinations of genotypes of ERCC2 K751Q, CD3EAP G-21A and XRCC3 T241M. The numbers at risk at 0, 24, 48 and 72 months are presented below the figure. (a) The variant haplotype of ERCC2 K751Q and CD3EAP G-21A (green line); the variant haplotype of ERCC2 K751Q and the wildtype of CD3EAP G-21A (orange line); the wildtype of ERCC2 K751Q and variant haplotype of CD3EAP G-21A (blue line); wildtype of ERCC2 K751Q and CD3EAP G-21A (red line). (b) The variant haplotype of ERCC2 K751Q and XRCC3 T241M (green line); the variant haplotype of ERCC2 K751Q and the wildtype of XRCC3 T241M (orange line); the wildtype of ERCC2 K751Q and the variant type of XRCC3 T241M (blue line); the wildtype of ERCC2 K751Q and XRCC3 T241M (red line). (c) The variant haplotype of CD3EAP G-21A and XRCC3 T241M (green line); the variant haplotype of CD3EAP G-21A and the wildtype of XRCC3 T241M (orange line); the wildtype of CD3EAP G-21A and the variant haplotype of XRCC3 T241M (blue line); the wildtype of CD3EAP G-21A and XRCC3 T241M (red line).

There was no association between ERCC2 K751Q, CD3EAP G-21A and XRCC3 T241M polymorphisms. The additive effect of the polymorphisms was tested by Cox regression analysis including multiplicative variables.

When ERCC2 K751Q and CD3EAP G-21A genotypes were combined, additive effects of the 2 genotypes were observed (Table V). Wildtype carriers had a median TTF of 23.4 month. Carriers of variant alleles of either ERCC2 K751Q or CD3EAP G-21A had a median TTF of 27.6 month (p = 0.019) and 28.6 month (p = 0.046), respectively, whereas carriers of variant alleles of both genotypes had a median TTF of 48.6 month (p < 0.001) (Fig. 3a.). When ERCC2 K751Q was combined with XRCC3 T241M, homozygous carriers of the wildtype alleles had a median TTF of 17.5 month (Table V). Carriers of the variant allele of ERCC2 K751Q had a median TTF of 27.6 month (p = 0.21) and carriers of the variant allele of XRCC3 T241M, 29.9 month (p = 0.21), whereas carriers of both alleles had a median TTF of 49.9 month (p = 0.0001) (Fig. 3b). This indicates that the combination of suboptimal repair of both double strand breaks and bulky adducts is associated with increased TTF. When CD3EAP G-21A was combined with XRCC3 T241M, wildtype carriers of both genotypes had a median TTF of 22.5 month. Carriers of the variant alleles of either CD3EAP G-21A or XRCC3 T241M had a median TTF of 69.4 month (p = 0.0001) and 45.4 month (p = 0.0004), respectively, whereas carriers of variant alleles of both genotypes had a median TTF of 46.8 month (p = 0.0006) (Fig. 3c). Thus, TTF was the same for all carriers of variant alleles. This indicates that no additional effects on TTF were obtained by combining the 2 genotypes and that only 1 variant is needed to obtain increased TTF. We therefore constructed a simple binary variable indicating whether the patients were carriers of the variant alleles of either CD3EAP G-21A or XRCC3 T241M or both, versus wildtype carriers. This variable had a HR at 0.5 and a p value < 0.0001 with TTF as a dependent variable.

Table V. Univariate Analysis of the Effects of Combinations of Polymorphisms on TTF

ERCC2 K751Q	CD3EAP G-21A	XRCC3 T241M	N	Median TTF (mo)	HR	p	Median OS (mo)	HR	p
1 Homozygous carrier of the wild type allele. 2 Not included in the analysis. 3 Heterozygous or homozygous carrier of the variant allele. 4 Values in parentheses are 95% CI. 5 Median survival not reached.
AA1	GG	NI2	94	23.4	1		52.0	1
AA	AG + AA3	NI	42	28.6	0.6 (0.4–1.0)4	0.046	–5	0.6 (0.3–1.0)	0.056
AC+ CC	GG	NI	128	27.6	0.7 (0.5–0.9)	0.019	65.7	0.7 (0.5–1.1)	0.11
AC + CC	AG + AA	NI	68	48.6	0.4 (0.3–0.6)	<0.0001	76.5	0.6 (0.3–0.9)	0.019
AA1	NI	CC	46	17.5	1		65.6	1
AA	NI	CT + TT	51	29.9	0.7 (0.4–1.2)	0.25	–5	0.8 (0.4–1.6)	0.68
AC+ CC	NI	CC	63	27.6	0.7 (0.5–1.2)	0.21	65.9	0.9 (0.5–1.6)	0.66
AC + CC	NI	CT + TT	81	49.9	0.4 (0.2–0.6)	0.0001	76.5	0.7 (0.4–1.2)	0.20
NI	GG	CC	78	22.5	1		65.6	1
NI	GG	CT + TT	80	45.4	0.5 (0.3–0.7)	0.0004	81.6	0.8 (0.5–1.2)	0.27
NI	AG + AA	CC	31	69.4	0.4 (0.2–0.7)	0.001	–5	0.6 (0.3–1.2)	0.11
NI	AG + AA	CT + TT	52	46.8	0.4 (0.3–0.7)	0.0006	–5	0.6 (0.3–1.0)	0.065

Multivariate analysis of prognostic markers

The polymorphisms were compared to known prognostic factors (β2-microglobulin, creatinine, albumin, sex, age and Durie–Salmon stage). No significant correlations were found (data not shown) except for the XRCC3 T241M polymorphism and age (high age associated to wild type XRCC3 T241M, p = 0.02).

The univariate significant polymorphisms were tested in a multivariate analysis to see whether they had independent prognostic value. They were tested separately against the statistically significant parameters from the univariate analyses to calculate adjusted hazard ratios and p values. β2-Microglobulin was not analyzed routinely in 1 center. Therefore, only 240 patients were available in the multivariate analysis with the CD3EAP G-21A and ERCC2 K751Q polymorphisms, and 160 patients were available with the XRCC3 T241M polymorphism.

The adjusted hazard ratios and p values were similar to the unadjusted hazard ratios (Table VI).

Table VI. Multivariate Analysis of Characteristic Affecting TTF

Parameter	TTF
	p value	Relative risk
1 β2-Microglobulin and creatinine were treated as continuous variables and log transformed. The hazard ratio indicates risk when values are doubled. 2 Values in parentheses are 95% CI. 3 The hazard ratios of progression for the variants were calculated with respect to having wildtype in CD3EAP G21A and XRCC3-T241M. The p value is for all combinations used as a categorical variable.
B2-Microglobulin1	0.001	1.4 (1.2–1.8)2
ERCC2 K751Q	0.009	0.5 (0.3–0.8)
CD3EAP G-21A or XRCC3 T241M3	0.0001	0.4 (0.2–0.6)

Using a wider range of commonly used prognostic variables (all shown in Table II except for ISS) and a backward stepwise method, all polymorphisms stayed in the model as statistically significant prognostic factors (data not shown).

We constructed a multivariate analysis including all the values from the univariate analysis.

The multivariate analysis showed that known prognostic markers such as stage according to Durie–Salmon and ISS were not prognostic markers for TTF. Elevated β2-microglobulin level was a prognostic marker for adverse TTF and OS. Creatinine levels correlated to poor OS. The polymorphisms ERCC2 K751Q, CD3EAP G-21A and XRCC3 T241M were significant prognostic markers for prolonged TTF, but only CD3EAP G-21A was a prognostic marker for prolonged OS.

Discussion

In this study we have explored the importance of genetic constitution for prognosis in a cohort of 348 patients treated with high-dose chemotherapy and ASCT. We report 3 new parameters related to genetic variations that influences TTF. In univariate analysis as well as Cox multivariate analysis, which included 146 patients (and 223 when β2-microglobulin and ISS were excluded), we found a statistically significant effect on TTF of ERCC2 K751Q, CD3EAP G-21A and XRCC3 T241M and on OS of CD3EAP G-21A. We have found that carriers of variant alleles of the polymorphisms ERCC2 K751Q and XRCC3 T241M had an increased TTF when compared to homozygous carriers of the wildtype allele. The 2 polymorphisms results in amino acid substitutions that probably modify the catalytic capacity of the DNA repair enzymes.28

Our interpretation of the results is that in treatment with high-dose alkylating chemotherapy it is an advantage to have low DNA repair capacity. With a low DNA repair capacity it is more likely that tumor cells are killed by the high-dose therapy, because unrepaired DNA damage will make the cells go into apoptosis. With a high DNA repair capacity, efficient DNA repair may allow tumor cells to escape apoptosis, eventually leading to relapse for the patient. Melphalan, which was used for bone marrow ablation, is an alkylating agent giving rise to bulky DNA adducts and DNA crosslinks.29XRCC3 and certain NER genes are known to be crucial for the repair of crosslinks and NER in general takes care of bulky adducts. Thus, it is a reasonable assumption that NER as well as double-strand–breakage repair plays a role in the cellular response to the drug.

This assumption is supported by our findings of the longest TTF of 49.9 month in the subgroup of patients with the alleles associated with suboptimal repair in both NER and double strand repair. No effect was however seen on survival. This may be explained by the use of new drugs such as IMiDs and proteasome inhibitors at progression of disease after ASTC or because less events of OS was registered. In contrast, patients with ineffective DNA repair may be at risk for secondary malignancies.

Few studies have focused on polymorphisms and the outcome of chemotherapy. Different chemotherapeutic agents vary in the amount and type of DNA damage induced and to which extent apoptosis is induced. Furthermore, the interplay between genetically determined factors, such as DNA repair efficiency and apoptosis threshold, on the one hand, and treatment regimen on the other, require large study groups of homogeneous populations and treatment strategies for evaluation of the relationship between polymorphisms and treatment outcome. In a large study of 341 elderly patients with acute leukemia, it was found that carriers of the variant homozygote of ERCC2 K751Q had the poorest prognosis and this genotype was associated with the highest risk of developing AML after chemotherapy.5 We selected a large group of patients that included the majority of the patients in Denmark who were treated in accordance to the treatment protocols of NMSG. In our setting, patients carrying the variant allele of ERCC2 K751Q had the best prognosis. Which DNA repair status is favorable may thus vary depending on the type of cancer and chemotherapeutic treatment strategies.

The polymorphism CD3EAP G-21A was associated with large differences in both TTF and OS, and 31% were carriers of the favorable variant allele. No previously published work has shown effect on treatment outcome in relation to polymorphisms in CD3EAP. The function of CD3EAP protein is unknown, but it colocalizes with the RNA polymerase I transcription initiation factor UBF/NOR-90 throughout all stages of the cell cycle.13 Its localization could suggest that the protein impinges on ribosome synthesis, one of the central features of cell proliferation. Involvement in the immune function of this gene product has been suggested.14 However, the studied polymorphism in CD3EAP does not result in amino acid substitutions and it is located close to the genes ERCC1 and PPP1R13L. The effect on TTF and OS was only observed for heterozygous carriers of CD3EAP G-21A. This indicates that it may not be the polymorphism itself that has biological impact but rather a polymorphism that cosegregates with it. It is therefore possible that the variant allele of CD3EAP G-21A is a marker for another genetic variation that modifies the function of ERCC1 or PPP1R13L and the polymorphism may therefore not be involved in the causal explanation of the prolonged TTF and OS. Our combination analysis suggests an independent function of CD3EAP G-21A and ERCC2 K751Q.

Our results indicate that polymorphism in CD3EAP G-21A and ERCC2 K751Q may have effects on TTF, and for ERCC2 K751Q also on OS, in women only. Data from the Danish Cancer Registry collected from 1989 to 2001, including all 3,500 patients with MM, indicate that women have a small but statistically significant longer survival when compared to men. Thus, women had a median OS of 26 months when compared to 22 months for men (p = 0.02).

There are several examples of interaction between gender and polymorphisms in this specific chromosomal region. Among women in the Danish Diet, Cancer and Health cohort consisting of 57,000 Danes, homozygous carriers of the haplotype ERCC1 N118N^A, CD3EAP G-21A^G and PPP1R13L IVS1 A4364G^A were at increased risk of breast cancer,17 an effect that is not found among males with cancer of the testis.19 In a large study of 1,092 lung cancer cases and 1,240 controls, a borderline statistically significant interaction was found between ERCC2 Asp312Asn and gender in relation to risk of lung cancer,30 and in a recent study of 428 lung cancer cases and 800 controls, the relation between risk of lung cancer and the haplotype ERCC1 N118N^A, CD3EAP G-21A^G and PPP1R13L IVS1 A4364G^A was statistically significantly different between the 2 genders and only female carriers were at increased risk. Furthermore, gene-smoking interaction with the haplotype was detected among women, but not among men.31

Abnormalities of chromosome 13 and translocations involving the IgH gene locus (14q32) have been associated with survival and response to high-dose treatment.32, 33 These analyses are related to the biology of the tumor cell and require purification of myeloma cells from bone marrow sample. In this study we present data that individuals with polymorphisms in ERCC2 K751Q, CD3EAP G-21A and XRCC3 T241M have prolonged TTF. These results may explain part of the differences in TTF noted among patients. Our results indicate that high-dose therapy may be beneficial for MM patients with defective DNA repair, while treatment improvements, such as dose escalation or inclusion of drugs known to inhibit DNA repair, may be relevant for carriers of the wildtype alleles.

Acknowledgements

Participants of NMSG. H. Knudsen, E. Gaarsdal, B. Ingemann, C. Lønskov and H. Larsen, all from KAS Herlev; S.T. Clausen and T. Nørgaard, both from Hillerød Sygehus; L. Drivsholm and P.P. Clausen, both from Næstved Sygehus; K. Bendix, A. Lønskov, A.Z. Kudsk and J.L. Nielsen, all from Århus Amtssygehus; M. Frederiksen, Haderslev Sygehus; B. Bengtsson and N. Korsgaard, both from Esbjerg Sygehus; E. Ludvigsen, Sønderborg Sygehus; C. Rygaard, H.S. Hvidovre Hospital; H.K. Thomsen and A. Kjøge, both from Bispebjerg Hospital; F. Egede, Sygehus Syd Svendborg; E. Worthington, Tønder Sygehus; O.V. Gadeberg, Vejle Sygehus; N. Abildsgård and N. Tinggaard, both from Odense Sygehus; H. Gregersen, Aalborg Sygehus; H. Kiær, Sygehus Fyn Svendborg; J.C. Andreassen and H.lK. Nielsen, both from Sygehus Vestsjælland; H. Sciolla, Rigshospitalet; A. Juhl, Nykøbing Falster Sygehus; B. á Steig, Landssjukrahusid, Faroe Islands; B. Bach, Viborg Sygehus. All these persons are from Denmark.