Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Clin Cancer Res. Author manuscript; available in PMC 2009 June 15.
Published in final edited form as:
PMCID: PMC2675883

Association of Megalin Genetic Polymorphisms with Prostate Cancer Risk and Prognosis



Megalin, an endocytic receptor expressed by prostate epithelial cells, can internalize biologically active androgens bound to sex hormone binding globulin. Genetic variation within megalin could potentially influence levels of steroid hormone uptake.

Experimental Design

Forty haplotype-tagging single nucleotide polymorphisms (htSNPs) were analyzed in a population-based, case-control study of 553 Caucasian men, who were diagnosed with prostate cancer between ages 40–64 from the Seattle-Puget Sound region, and 534 control men. Prostate cancer risk was estimated using adjusted unconditional logistic regression for both individual SNPs and haplotypes. Risks of disease recurrence/progression and prostate-specific cancer mortality were estimated using Cox proportional hazards (PH) regression.


We found no strong evidence of altered risk of developing prostate cancer for any of the htSNPs when they were assessed individually or in haplotypes. However, three htSNPs were significantly associated with both disease recurrence/progression and mortality. Risk of recurrence/progression alone was also associated with five additional htSNPs and six other htSNPS showed evidence of modification by primary androgen deprivation therapy (ADT). Two additional htSNPs were significantly associated with altered risk of death from prostate cancer.


Preliminary results suggest that common genetic variation within the megalin gene could alter both risk of recurrence/progression and prostate-specific cancer mortality. In addition, ADT effectiveness may be modified by the activity of this gene. To our knowledge, this is the first study that has examined polymorphisms within the megalin gene for associations with prostate cancer risk and outcomes.

Keywords: prostate cancer, polymorphisms, megalin, case-control, outcomes


Androgens are believed to contribute to both the etiology and progression of prostate cancer1. Previously, it was thought that the only way the androgen hormone entered the cell and became biologically active was after disassociation from its plasma carrier protein, sex hormone binding globulin (SHBG), and subsequent internalization of the lipophilic compound via free diffusion 2. A recent study by Hammes et al., showed that megalin, a member of the LDL receptor gene family located on chromosome 2q24-q31, may play a role in the internalization of androgen into the prostatic cell 3. These researchers showed that megalin, a multiligand endocytic receptor, carries a specific binding site for androgen still bound to SHBG, and serves as an active uptake mechanism of the entire steroid hormone complex. Megalin’s role in the uptake of androgen in the mature healthy human prostate cell is not known, but our preliminary immunohistochemical assay studies of megalin gene expression have found increased levels of the protein in malignant relative to benign prostate cells (P.S. Nelson and E. Mostaghel, unpublished data).

Polymorphisms within the megalin gene (LRP2, low density lipoprotein-related protein 2) could potentially alter the activity of the encoded protein to either promote or inhibit active transport of androgen into the prostate cell resulting in a phenotype that has altered levels of androgen in the prostate cell. Whether this phenotype would have more of an impact on normal versus malignant prostate cells is difficult to predict. If megalin does in fact play a significant role in the internalization of androgen in normal mature prostate cells, then genetic variation could potentially affect the risk of developing prostate cancer. It is also plausible that prostate cancer cells may benefit from increased internalization of androgens via megalin’s active transport mechanism 4. Indeed, our immunohistochemical assay evidence does support a role of megalin in tumor progression. This lends credence to the hypothesis that a megalin phenotype that alters the uptake of androgen could modify the progression from a localized androgen-dependant state into the androgen-independent and metastatic forms of prostate cancer. This association could be observed in either the risks of clinically diagnosed aggressive forms of prostate cancer or by disease outcomes. In addition, changes in protein function mediated by genetic variants could theoretically modify the reaction of prostate cells to androgen-deprivation therapy (ADT). The goal of this study was to measure common genetic variation within megalin and to explore potential associations with disease risk, recurrence/ progression, prostate-specific mortality, and possible interactions with primary ADT in a population-based, case-control study.

Materials and Methods

Study Population

Study subjects were enrolled in a population-based prostate cancer case-control study that has been described previously 5. Prostate cancer cases were identified from the metropolitan Seattle-Puget Sound population-based tumor registry that is operated as part of the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program. The SEER registry also provided information on tumor characteristics, primary therapy, and vital status. Underlying cause of death was confirmed from death certificates. Eligible individuals were Caucasian or African American men aged 40–64 years who were newly diagnosed with histologically confirmed prostate cancer between January 1, 1993 and December 31, 1996. Of the 916 men eligible for the study, 752 completed structured in-person interviews (82.1%). We obtained peripheral blood leukocyte samples yielding sufficient DNA for genotyping from 585 (77.8%) of the interviewed cases. Analysis was further limited to Caucasian men bring the total cases to 553 (94.5%).

Controls were identified through random digit telephone dialing (RDD) using a clustering factor of five residences per sampling unit. Controls were frequency matched to cases by age (same 5-year group). Eligible controls were Caucasian or African American men aged 40–64 at the reference date, with no history of prostate cancer. Complete household census information was obtained for 94% of the 21,116 residential telephone numbers contacted. Of the 941 men who were identified and agreed to participate, 703 were interviewed (74.7%) making the overall level of participation 70.2% (94% × 74.7%). From these potential controls our analysis was limited to the 549 (78.1%) from whom we had sufficient DNA for genotyping. Analysis was further restricted to Caucasian men bring the total number of controls to 534 (97%).

Cases and controls completed detailed in-person interviews covering demographic, medical, lifestyle, and family cancer history conducted by trained male interviewers. Blood specimens were processed and DNA isolated using standard protocols 6. There were no differences between the non-genotyped and genotyped cases and controls with respect to age, race, family history of prostate cancer, or clinical characteristics of the patients. In January 2004 a self-administered follow-up questionnaire collecting information on quality of life, secondary therapies, follow-up PSA results, and prostate cancer recurrence/progression was sent to the cases. Of the 630 cases who were alive and previously consented to future contact, the survey was completed by 520 (82.4%). Out of these respondents, 405 (77.9%) had DNA and were Caucasian. There were no differences between follow-up survey respondents versus non-respondents with respect to tumor characteristics or primary therapy. This study was approved by Fred Hutchinson Cancer Research Center’s Institutional Review Board and genotyping was approved by the Internal Review Board of the National Human Genome Research Institute.

SNP Selection and Genotyping

SNPs that captured the genetic variability in the gene were selected using publicly available data from the HapMap consortium1 from a 248 kb region that comprised the LRP2 locus and 13 kb of the 5’ end/promoter (from 169,804,367 to 170,048,683 in NCBI build 35). There were 219 SNPs with minor allele frequencies greater than 5% in the Caucasian HapMap dataset. Haploview software version 3.11 was used to generate pairwise LD estimates and define haplotype blocks using criteria outlined first by Gabriel et al. then by the Solid Spline approach 7,8. The user defined linkage disequilibrium criteria for defining the blocks was fraction of informative pairwise comparisons ≥ 0.95 and D’ > 0.8. This excludes SNPs with a minor allele frequency less than 5% by default. A total of the 40 haplotype tagging SNPs (htSNPs) were then selected to capture the variation within each block and then regions in between block 9. We defined common haplotypes as those with a frequency of greater than 5%. The set of optimal htSNPs included 3 non-synonymous SNPs that were forced into the htSNP selection.

The Applied Biosystems (ABI) SNPlex™ Genotyping System was used to genotype SNPs in individual DNA samples. Proprietary GeneMapper® software was used for calling alleles2. Discrimination of the specific SNP allele is carried out with the ABI 3730xl DNA Analyzer and is based on the presence of a unique sequence assigned to the original allele-specific oligonucleotide.

Quality control included genotyping of 76 blind duplicate samples, which revealed 99% agreement on genotyping calls across all SNPs assayed. In addition, each batch of DNA aliquots genotyped incorporated similar numbers of case and control samples, and laboratory personnel were blinded to the case-control status of samples. The overall call rate was 99.0%. Subjects had to have at least 75%, or 30 out of the 40 SNPs, successfully genotyped to be included in the study. Of the 56 case specimens included who did not have completely successful genotyping, 34 had only one SNP with unknown genotype, 12 had two SNPs with unknown genotype, and 10 had three to nine SNPs with unknown genotypes. Of the 30 controls did not have complete genotyping, 16 had only one SNP with unknown genotype, and 6 had two SNPs with unknown genotype, and 8 had three to nine SNPs with unknown genotypes.

Statistical Analysis

SNP genotype frequencies were examined for Hardy-Weinberg equilibrium (HWE) using the χ2 statistic. Data were analyzed using unconditional logistic regression to calculate odds ratios as estimates of the relative risk of prostate cancer associated with any given individual htSNP. For each htSNP we classified carriers of the variant or less common allele as the exposure group and then used both dominant and co-dominant models, excepting the polymorphic loci for which no individuals were homozygous for the variant genotype. We included the design variable of age in all regression models. We also assessed possible confounding effects of first degree family history of prostate cancer, PSA testing history (none versus any PSA test performed in the previous 5 years), history of benign prostatic hyperplasia (none, diagnosis ≤ 2 years before reference date, diagnosis >2 years before reference date), and body mass index (BMI, <25, 25–29, ≥ 30 kg/m2). As adjustment for none of these variables appreciably altered risk estimates, we did not include them as covariates. Trend tests were used to assess gene dosage and a global p values were estimated to assess overall gene effect. Associations with individual htSNPs according to Gleason score [2–6 or 7 (3+4) vs. 7(4+3) or 8–10], tumor stage (local vs. regional/distant) and a composite prostate cancer aggressiveness score were examined using polytomous regression. Classification parameters of the composite “non-aggressive” phenotype included diagnosis at local stage, Gleason score of 2–6 or 7 (3+4), and serum PSA less than 20 ng/ml whereas the “aggressive” phenotype included diagnosis at regional/distant stage, Gleason score 7(4+3) or 8–10, or serum PSA 20 ng/ml or greater.

Because the htSNPs were chosen to represent haplotype variation it is necessary to include both individual SNP and haplotype analysis for a complete assessment of genetic variation with disease risk. Two haplotype analyses were carried out. In the first, haplotype risk was assessed within the each block using criteria defined by Gabriel et al 7. Using data from control subjects, Haploview software version 3.2 was used to estimate pairwise linkage disequilibrium (LD) between the htSNPs and define haplotype blocks. In the second analysis, haplotype risk was assessed for a haplotype containing only htSNPs which demonstrated an individual significant association with prostate cancer risk. Risk was assessed using Haplostat software (version 1.2.13) for R (version 2.1.04), which employs the expectation-maximization (EM) algorithm to estimate haplotype frequencies and an iterative two-step EM model to estimate the association between individual haplotypes and risk assuming an additive model 10. Overall differences in risk of prostate cancer across haplotypes were assessed using global score test adjusted for age 11. Only haplotypes with estimated frequencies of >5% were included.

To examine associations between individual htSNPs with prostate cancer recurrence/progression and prostate cancer mortality we used Cox proportional hazards (PH) regression models adjusting for age, Gleason score, stage at diagnosis, and diagnostic prostate-specific antigen level. For estimating risk in the recurrence/progression model, the time-dependant variable was defined as time from diagnosis to the first self-reported evidence of development of metastasis. This variable was calculated using the data from the self-administered follow-up questionnaire as described previously 12. The censoring date for cases without a metastasis event was the date that their follow-up questionnaire was returned. Cases who had an initial diagnosis of metastatic disease were excluded from the analysis (n=5) as were cases who were alive and did not fill out a follow-up questionnaire (n=83). To account for cases diagnosed with localized or regional disease who died of metastatic prostate cancer before the follow-up survey was administered, the PH model was weighted. The weights were the inverse of the probability of having a metastasis date for men with a metastatic event and 1.0 for the men who were censored 13. For calculation of prostate-specific mortality, the time-dependant variable was defined as the time from diagnosis to death or censoring for men who remained alive. Living cases were censored on the date of the most recent match with the cancer registry, December 1st, 2006. Cases that died from other causes were censored at the time of death.

We also assessed the possibility of an interaction between genotype and ADT with both disease recurrence/progression and survival. Using both data from the SEER registry and the self-report questionnaire, we defined cases who received androgen deprivation shots, pills, and/or orchiectomy within 12 months of initial diagnosis as having received primary ADT. Due to small numbers of cases receiving primary ADT overall (n=94) we limited both our mortality and recurrence/progression PH models to risk estimates assuming a dominant genetic model, combining heterozygotes and homozygous variants as the comparison group. Models with and without interaction terms were compared using the likelihood ratio test.

In an effort to account for multiple comparisons, the false discovery rate (FDR) was used to calculate corrected critical p-value to evaluate the robustness of all risk estimates 14. After consideration of SNP genotypes individually, all significant SNPs were included in a stepwise selection model using Akaike's Information Criterion to select the most parsimonious model15. SNPs that remained significant (p ≤ 0.05) after adjustment for each other, were included in the final model16. Both forward and backward selection models were compared, with equivalent results. All analyses were done using the STATA statistical package (version 9.2, STATA Corp., College Station, TX).


Cases and controls were similar in age (mean in cases, 57 years; in controls, 57 years). Cases were more likely to report a family history of prostate cancer, history of benign prostatic hyperplasia (BPH), and history of PSA tests (Table 1). The majority of prostate cancers were local stage tumors with low/moderate Gleason scores. All SNPs included in this study were statistically consistent (P > 0.05) with the Hardy-Weinberg equilibrium (HWE).

Table 1
Selected characteristics of Caucasian prostate cancer cases and controls.

There was no strong evidence of altered risk of developing prostate cancer for any of the htSNPs evaluated (Table 2). Two SNPs (rs3944004, rs2268373) showed slightly lower relative risks of prostate cancer for the homozygous variant carriers versus homozygous wildtype carriers. Two SNPs (rs2239598, rs831003) showed slightly increased relative risks for the heterozygote carriers as compared to homozygous wildtype carriers. However, the P for trend for all four of these SNPs was >0.05 and after adjustment for multiple comparisons none of these associations remained significant. When we assessed risk estimates by measures of tumor aggressiveness (Gleason score, stage, and composite score) we found no differences in risk estimates for any of the polymorphisms examined.

Table 2
Genotype distribution and odds ratios (95% CI) for the association between megalin htSNPs and prostate cancer risk.

Within the forty htSNPs included in this study, we defined 8 haplotype blocks (A-H, see Table 2) ranging from 2 to 5 htSNPs. Thirteen htSNPs did not fall into any haplotype block structure. We found no evidence of altered relative risks of prostate cancer for any of the haplotypes in any of the blocks. The four SNPs that showed slightly altered risks at the individual level (rs3944004, rs2268373, rs2239595, and rs831003), displayed no evidence of altered risks for prostate cancer when a haplotype containing only these four htSNPs was considered.

There were 86 cases with physician-diagnosed recurrence/progression, with an average 8.8 years of follow-up (range 0.2–11.9 years) after diagnosis. Of these 86 cases, 67 (78%) were diagnosed at low/moderate Gleason score, 55 (64%) were diagnosed at a localized stage, and 14 (16%) were treated with primary ADT. Eight htSNPs showed altered risks of disease recurrence/progression overall (Table 3). When these SNPs were adjusted for multiple comparisons, only three SNPs (rs830994, rs830995, and 831003), all within the same haplotype block “E”, retained significance. We then considered all eight SNPs and then only the three block “E” SNPs using stepwise regression and found that only rs830994 remained significant at p = 0.001, independent of all other SNPs in the model. An additional six htSNPs, while not demonstrating altered risks for all cases combined, did show some evidence of different risks of recurrence/progression by ADT (Table 3).

Table 3
Association of megalin htSNPs genotypes with prostate cancer recurrence/progression by primary androgen deprivation therapy (ADT).

There were 33 cases who died of prostate cancer in the average 10.4 years of follow-up (range 1.3–13.4 years). Of these 33 cases, 12 (36%) were diagnosed at low/moderate Gleason score (2–6 or 7 (3+4)), 6 (18%) were diagnosed at a localized stage, and 26 (79%) were treated with primary ADT. Five htSNPs were significantly associated with altered risks of death from prostate cancer, with no evidence of interaction with ADT (Table 4). None of these 5 SNPs remained significant after adjustment for multiple comparisons. Using stepwise regression, we observed independence for only two SNPs, rs2300446 at p = 0.02 and rs830094 at p = 0.03. The htSNP rs830994 was found to have genotypes that had altered risks of both recurrence/progression and mortality independent of all other SNPs under the uncorrected multiple comparison models.

Table 4
Association of megalin htSNPs genotypes with prostate cancer-specific mortality by primary androgen deprivation therapy (ADT).


Because of the apparent etiologic role for androgens in the development of prostate cancer, several candidate gene studies have focused on genetic polymorphisms found in the cascade of enzymes involved in androgen metabolic pathway. This is one of the first epidemiologic studies to examine megalin, a gene that functions to actively transport androgen into the prostatic cell. Results from this study do not support a relationship between megalin polymorphisms and the risk of developing prostate cancer. The megalin-deficient mouse model described by Hammes et al. demonstrated the importance of megalin in the developmental stage by the impaired descent of testes in males, a phenotype similar to animals treated with androgen-receptor antagonists. However, the importance of megalin as an active uptake mechanism in the mature prostatic cell is unknown and under normal physiologic conditions, the primary means of androgen internalization is likely to be via free diffusion across the cell membrane. Thus, under most conditions, megalin may not substantially influence the development of prostate cancer.

While megalin may not play an important role for androgen internalization in normal prostate tissue, it may serve as a means for the malignant cell to increase androgen uptake. Using immunohistochemical assays, we have determined that neoplastic prostate epithelium expresses greater levels of megalin protein than adjacent benign epithelium (P.S. Nelson and E. Mostaghel, unpublished data). We hypothesize that polymorphisms in the megalin gene that result in altered transport activity may influence the outcome of the disease. Our preliminary results suggest that genetic variants within the megalin gene may alter both risk of prostate cancer-specific death and disease recurrence/progression.

ADT takes advantage of the dependence of the prostate cell on androgen to temporarily arrest further malignant growth. This approach is not curative, and with time the cells will become resistant to ADT. Although this phase is termed “androgen-independence”, it has been shown that the androgen receptor retains activity as demonstrated by the continued expression of androgen regulated genes such as PSA. If megalin provides an important secondary means of steroid uptake to circumvent the efficiency of ADT and has a true altered activity phenotype, outcomes could be modified by influences between genotypic variation within megalin and the physiological pressures of ADT. Although we were limited by power, our results do suggest a potential interaction at least with disease recurrence/progression for six of the megalin htSNPs tested.

This was a population-based study with detailed information allowing for analysis of potential interaction and confounders. While this study had enough power to detect modest relative risks for prostate cancer at the individual SNP level, we were limited by sample size in analyses of outcomes and in subset analyses. In addition, our Caucasian-only population may make our findings less generalizable. It is possible that some control men may have had undiagnosed prostatic cancer, leading to misclassification of disease and attenuation of the risk estimates toward the null. However, over 86% of the controls in this study reported that they had previously had a DRE examination and/or PSA test for the detection of prostate cancer. In addition, we measured PSA in the serum of in a random sample of 400 controls within our study and found only 39 (9.8%) men with a PSA >=4.0 ng/ml and only 6 of these men (1.5%) with a PSA >=10.0 ng/ml. Because of the large size of the megalin gene and the lack of any previously published findings on functional polymorphisms, we elected to use the haplotype tagging approach. This was a cost effective means to identify susceptibility alleles across the entire gene region.

In conclusion, while this study does not support a role between megalin polymorphisms and the risk of developing prostate cancer, several polymorphisms appear to play a role in the prostate cancer outcomes following diagnosis. Both the htSNPs identified in this study and SNPs in linkage disequilibrium with these htSNPS need to be examined to determine if, and how, they affect megalin protein function. A true test for the robustness of our findings will be replication in a larger dataset. In addition, a larger study may be able to expand on the initial findings here, suggesting a role for megalin polymorphisms and outcomes associated with use of ADT.



This work was supported by grants RO1-CA56678, RO1-CA82664 and P50-CA97186 from the National Cancer Institute; additional support was provided by the Fred Hutchinson Cancer Research Center, and the Intramural Program of the National Human Genome Research Institute.

Reference List

1. Platz EA, Giovannucci E. Prostate cancer. In: Schottenfeld D, Fraumeni J, editors. Cancer epidemiology and prevention. Oxford; New York: Oxford University Press; 2006. pp. 1128–1150.
2. Mendel CM. The Free Hormone Hypothesis: a Physiologically Based Mathematical Model. Endocrine Reviews. 1989;10:232–274. [PubMed]
3. Hammes A, Andreassen TK, Spoelgen R, et al. Role of endocytosis in cellular uptake of sex steroids. Cell. 2005;122:751–762. [PubMed]
4. Andreassen TK. The role of plasma-binding proteins in the cellular uptake of lipophilic vitamins and steroids. Hormone and Metabolic Research. 2006;38:279–290. [PubMed]
5. Stanford JL, Wicklund KG, McKnight B, et al. Vasectomy and risk of prostate cancer. Cancer Epidemiology, Biomarkers and Prevention. 1999;8:881–886. [PubMed]
6. Sambrook J, Fritsch EF, Maniatis T. Isolation of high-molecular weight DNA from mammalian cells. In: Nolan C, editor. Molecular Cloning: A Laboratory Manual. 1989. 9.16–9.19.
7. Gabriel SB, Schaffner SF, Nguyen H, et al. The structure of haplotype blocks in the human genome. Science. 2002;296:2225–2229. [PubMed]
8. Barrett JC, Fry B, Maller J, et al. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005;21:263–265. [PubMed]
9. de Bakker PI, Yelensky R, Pe'er I, et al. Efficiency and power in genetic association studies. Nature Genetics. 2005;37:1217–1223. [PubMed]
10. Schaid DJ, Rowland CM. Robust transmission regression models for linkage and association. Genetic Epidemiology. 2000;19(Suppl 1):S78–S84. [PubMed]
11. Schaid DJ, Rowland CM, Tines DE, et al. Score Tests for Association Between Traits and Haplotypes When Linkage Phase Is Ambiguous. American Journal of Human Genetics. 2002;70:425–434. [PubMed]
12. Agalliu I, Lin DW, Salinas CA, et al. Polymorphisms in the glutathione s-transferase M1, T1, and P1 genes and prostate cancer prognosis. Prostate. 2006;66:1535–1541. [PubMed]
13. Robins JM, Rotnitzky A, Zhao LP. Estimation of regression coefficients when some regressors are not always observed. Journal of the American Statistical Association. 1994;89:846–866.
14. Benjamini Y, Drai D, Elmer G, et al. Controlling the False Discovery Rate in Behavior Genetics Research. Behavioural Brain Research. 2001;125:279–284. [PubMed]
15. Li W, Nyholt DR. Marker selection by Akaike information criterion and Bayesian information criterion. Genetic Epidemiology. 2001;21(Suppl 1):S272–S277. [PubMed]
16. Cordell HJ, Clayton DG. A Unified Stepwise Regression Procedure for Evaluating the Relative Effects of Polymorphisms Within a Gene Using Case/Control or Family Data: Application to Hla in Type 1 Diabetes. American Journal of Human Genetics. 2002;70:124–141. [PubMed]