Insulin-like growth factor-1 genotypes and haplotypes influence the survival of prostate cancer patients with bone metastasis at initial diagnosis

From July 1980 to September 2008, 215 native Japanese patients of prostate cancer with bone metastasis at initial presentation in Akita University Hospital and its related community hospitals, Kyoto University and Tohoku University Hospitals were enrolled in this study. Pathological diagnosis was made by prostate needle biopsy specimens and metastasis was identified by X-rays, CT scans, or bone scintigraphy. All the patients had no previous treatments at presentation and underwent surgical castration or luteinizing hormone-releasing hormone (LH-RH) analogues with or without antiandrogens as primary endocrine therapy. When the treatment failure was observed, optional therapies, including other antiandrogens, estrogens, steroids, chemotherapeutic agents, palliative radiation, or a combination of these was added or replaced.

Pathological grading of needle biopsy specimens was performed according to Gleason grading system by local pathologists with no designated primary pathologist. In 10 patients, the final pathological grade was not determined because no grade information was described in the final report or a different grading system was applied by the local pathologists. Pretreatment HGB, ALP, lactate dehydrogenase (LDH), and PSA levels before the initial treatment of prostate cancer were obtained from medical charts. An independent end-point reviewer in each institution determined the cause of death on the basis of standardized extractions from the patients’ medical files without providing genotype data of each patient.

Written informed consent was obtained from all the patients enrolled in this study for the use of their DNA and clinical information. This study was approved by the Institutional Review Board (the Ethical Committee) in each institution.

DNA was extracted from a peripheral blood sample of each patient using a QIAamp Blood Kit (QIAGEN, Hilden, Germany) or standard phenol-chloroform. We divided the IGF-1 gene into three LD blocks according to a previous report by Johansson et al. (Figure 1) [25]. The representative polymorphisms in each LD block were chosen with reference to literatures as the genes previously described to be associated with the circulating IGF-1 level or the increased risk of prostate cancer [27] and SNP database of international HapMap project (http://​hapmap.​ncbi.​nlm.​nih.​gov/​) [28]. The CA repeat polymorphism in the promoter region and rs12423791 [GenBank] were selected from representative polymorphisms in the LD block 1 and 2, respectively. rs6220 [GenBank] and rs7136446 [GenBank] were selected as tag SNPs for a haplotype analysis in the LD block3. Genotypes of the CA repeat polymorphism were determined by an automated sequencer (ABI PRISM 310 Genetic Analyzer) with GENESCAN software (Applied Biosystems, Foster City, CA) as described previously [17]. Other three SNPs were genotyped using polymerase chain reaction (PCR)-restriction fragment length polymorphism (RFLP) method. The sequence of forward and reverse primers used for genotyping of the SNPs were as follows; 5^′-GCTGCTTCTTCCAATGAGAG-3^′ and 5^′-GAAAAGCATGTTGCTGCCTC-3^′ for rs12423791 (123 bp), and TGCCTAGAAAAGAAGGAATC-3^′ and 5^′-TGACTCTTCTATGCAGTTAC-3^′ for rs6220 (105 bp), and 5^′-CTTCTTGCAGAACTAAGCTCAAGTC-3^′ and 5^′-GCCTATTCATTTTCACATACTACCC-3^′ for rs7136446 (126 bp). Each PCR product was digested with DdeI, MspI, and MnlI, respectively, overnight at 37°C, and electrophoresed on 3.0% agarose gels to determine the genotype. Several samples were directly sequenced using Dye Terminator Sequencing Kit version 1.0 (PE Applied Biosystems) on an ABI prism 310 auto-sequencer to confirm the results of PCR-RFLP for each polymorphism.

The endpoint of this study was prostate cancer-specific survival. The survival time was calculated from the date of prostate cancer diagnosis to the date of death or the last contact with patients. To compare the survival, patients were dichotomized by the median value of age and PSA, by normal limits in HGB, ALP, LDH, by the tumor grade system (Gleason score). Haplotypes of the LD block 3 defined two SNPs, rs6220 and rs7136446, were inferred using expectation-maximization algorithm in SNPAlyze software ver.7 (Dynacom Co. Ltd., Chiba, Japan). Differences in survival between groups were analyzed using the logrank test. For the CA repeat polymorphism, the number of repeats was dichotomized as having or not having the 19-repeat allele. Each SNP or haplotype was evaluated using dominant, recessive, and addictive model, and the most statistically significant model was selected. The IGF-1 polymorphisms and clinicopathological prognostic factors were assessed by the Cox proportional hazard regression models. Age, Gleason score, PSA, HGB, ALP, LDH, and the LD block 3 haplotype were employed as a variable set in a full-variable model of multivariate analysis. Of the variables, Gleason score, HGB, ALP, and the LD block 3 haplotype were selected in a reduced variable model. Cancer-specific survival was estimated using the Kaplan–Meier method. All the statistical analyses were performed using SPSS software version 19.0 (IBM Japan Ltd., Tokyo, Japan) and two-sided P values of less than 0.05 were considered to indicate statistical significance.

The mean age (± SD) of the 215 patients was 70.2 ± 8.4 years (range, 45–89; median, 72 years). The mean follow-up period was 46.4 ± 36.1 months (range, 1–209; median, 37 months). Of 215 patients, bone metastasis alone, additional lymph nodes metastasis, and other visceral metastasis were seen at initial diagnosis in 106 (49.3%), 98 (45.6%), and 15 (7.0%), respectively. The distribution of Gleason score of biopsy specimen was < 7 in 15 patients (7.0%), 7–8 in 81 (37.7%), 9–10 in 109 (50.7%), and unknown in 10 (4.6%). Pretreatment PSA, HGB, ALP, and LDH levels are shown in Table 1. All the patients were received an endocrine therapy as an initial treatment, surgical castration alone in 29 (13.5%), LH-RH analogue alone in 53 (24.6%), combined androgen blockade in 131 (60.9%). Among 188 patients with available data, 91patinets (48.4%) achieved a PSA nadir less than 1 ng/ml, while 97 patients (51.6%) did not reach the level after initial endocrine therapies.

The repeat number of the CA repeat polymorphism ranged from 13 to 20, and 20 genotypes were observed. The distributions of the genotypes of the CA repeat polymorphisms, rs12423791, rs6220, and rs7136446 were shown in Table 2. Estimated haplotype frequencies of the LD block 3 (rs6220 - rs7136446) were 55.2% (T-T), 26.7% (C-T), 16.8% (C-C), and 1.3% (T-C). Thirty-seven patients with heterozygous genotype of both rs6220 and rs7136446 were estimated as having the T-T and C-C haplotypes because estimated haplotype frequency of the T-C haplotype was quite low. Haplotypes in all other patients were uniquely determined.

Kaplan-Meier curves demonstrated that patients with 19-repeat allele, C allele of rs12423791, or C allele of rs6220 had a significantly worse survival (P = 0.013, 0.014, or 0.014, respectively). Whereas, rs7136446 was not associated with patients’ survival (P = 0.371). Patients with at least one C-T haplotype showed significantly worse survival compared with those who had no C-T haplotype (P = 0.0003) (Figure 2). When the number of the genetic risk factors (19-repeat allele, C allele of rs12423791, or C-T haplotype) was considered, cancer-specific survival significantly shortened with increased the number of genetic risk factors (P = 0.002), and patients with all the genetic risk factors had significantly shorter survival than those with 0–2 risk factors (P = 0.0003) (Figure 3).

An univariate Cox proportional hazard regression analysis showed that cancer-specific survival was significantly lower in patients with Gleason score of 9 or higher (HR: 1.759, 95% CI: 1.151-2.687, P = 0.009), HGB less than 11.5 g/dl (HR: 2.251, 95% CI: 1.261-4.019, P = 0.006), ALP of 350 IU/ml or higher (HR: 2.836, 95% CI: 1.756-4.578, P = 0.00002), or LDH of 500 IU/ml or higher (HR: 2.638, 95% CI: 1.442-4.829, P = 0.002) (Table 3). Meanwhile, neither dichotomized age nor PSA was associated with cancer-specific survival. In a multivariate analysis including all the clinicopathological variables and haplotype of the LD block 3 as a representative genetic variable, higher Gleason score (HR: 1.766, 95% CI: 1.052-2.966, P = 0.031), higher ALP (HR: 2.598, 95% CI: 1.483-4.551, P =0.0008), and the C-T haplotype (HR: 2.619, 95% CI: 1.559-4.399, P = 0.0003) were independent factors predicting cancer-specific survival. HGB and LDH showed borderline significance (P = 0.061 and 0.059, respectively). In a reduced variable model, higher ALP (HR: 2.819, 95% CI: 1.695-4.689, P = 0.00007) and C-T haplotype (HR: 2.626, 95% CI: 1.603-4.305, P = 0.0001) were stronger independent predictors for the survival, followed by HGB (HR: 2.082, 95% CI 1.113-3.897, P = 0.022) and Gleason score (HR: 1.709, 95% CI: 1.054-2.771, P = 0.030) (Table 3).

Because the Gleason score and the pretreatment ALP level were shown to be significant prognostic factors along with the LD Block 3 haplotype by multivariate analysis (Table 3), we performed subgroup analyses according to the dichotomized Gleason score or the dichotomized pretreatment ALP level to compare survivals by presence or absence of C-T haplotype. Among patients with Gleason score of 9–10 (n = 108), those with C-T haplotype showed significantly worse survival than those having no C-T haplotype (P = 0.0002), while there was no significant difference (P = 0.365) in patients with Gleason score less than 9 (n = 96). Regarding ALP, patients with C-T haplotype showed significantly shorter survival than those with no C-T haplotype in either subgroup (ALP higher or lower than 350 IU/ml) (P = 0.010 or 0.009, respectively) (Figure 4).