Common Sequence Variants of the Macrophage Scavenger Receptor 1 Gene Are Associated with Prostate Cancer Risk

Abstract

Rare germline mutations of macrophage scavenger receptor 1 (MSR1) gene were reported to be associated with prostate cancer risk in families with hereditary prostate cancer (HPC) and in patients with non-HPC (Xu et al. 2002). To further evaluate the role of MSR1 in prostate cancer susceptibility, at Johns Hopkins Hospital, we studied five common variants of MSR1 in 301 patients with non-HPC who underwent prostate cancer treatment and in 250 control subjects who participated in prostate cancer–screening programs and had normal digital rectal examination and PSA levels (<4 ng/ml). Significantly different allele frequencies between case subjects and control subjects were observed for each of the five variants (P value range .01–.04). Haplotype analyses provided consistent findings, with a significant difference in the haplotype frequencies from a global score test (P=.01). Because the haplotype that is associated with the increased risk for prostate cancer did not harbor any of the known rare mutations, it appears that the observed association of common variants and prostate cancer risk are independent of the effect of the known rare mutations. These results consistently suggest that MSR1 may play an important role in prostate carcinogenesis.

The macrophage scavenger receptor 1 (MSR1 [MIM 153622]) gene was recently identified as a candidate gene for prostate cancer susceptibility, from a systematic search for prostate cancer (MIM 176807) genes at 8p, using multiple approaches such as linkage, direct sequencing, and association studies (Xu et al. 2002). Analysis of the MSR1 gene sequence in members of families with hereditary prostate cancer (HPC) identified six rare missense mutations (Pro36Ala, Ser41Tyr, Val113Ala, Asp174Tyr, Gly369Ser, and His441Arg) and one nonsense mutation (Arg293X). A family-based linkage and association test provided statistical evidence that these mutations cosegregate with prostate cancer (P=.0007). Further examination of these mutations in a collection of patients with non-HPC and unaffected men revealed that they were either not observed or were observed less frequently in men without prostate cancer. For white men, the rare MSR1 mutations were detected in 4.4% of cases, compared to 0.8% in unaffected men (P=.009); for African American men, these values were 12.50% and 1.82%, respectively (P=.01). These results provide genetic evidence that MSR1 may play an important role in prostate cancer susceptibility in both African American men and men of European descent.

The MSR1 protein, a Class A scavenger receptor, is a multidomain trimeric molecule composed of identical protein chains. It has two functional isoforms (Type I and Type II) and one nonfunctional isoform (Type III), generated by alternative splicing of a single 11-exon mRNA (Kodama et al. 1990; Emi et al. 1993). This macrophage-specific receptor is capable of binding a highly diverse array of polyanionic ligands, ranging from gram negative and positive bacteria and oxidized LDL to silica, and, correspondingly, has been linked to a wide variety of normal and pathological processes, including inflammation, innate and adaptive immunity, oxidative stress, and apoptosis (Platt and Gordon 2001). Although the exact role of MSR1 in prostate carcinogenesis is unknown, some or all of these processes have been implicated in the development of prostate cancer (De Marzo et al. 1999; Nelson et al. 2001). Recent findings showing that the degree of macrophage infiltration is associated with prostate cancer prognosis strengthen the link between MSR1 and prostate cancer (Lissbrant et al. 2000; Shimura et al. 2000).

Besides the seven rare MSR1 mutations we reported elsewhere (Xu et al. 2002), we identified four additional common sequence variants (>10%) after sequencing the PCR products of all 11 exons, exon-intron junctions, promoter region, and 5′ and 3′ UTRs of MSR1 in the germline DNA of probands from each of the 159 families with HPC (i.e., at least three first-degree relatives affected with prostate cancer). Together with the common missense change we identified elsewhere (Xu et al. 2002), there are five common sequence variants in our sequenced regions.

To evaluate the role of MSR1 sequence variants in prostate cancer susceptibility as it relates to the general population, we focused this study on the common sequence variants and patients with prostate cancer who do not meet criteria for HPC. Because of the limited number of African Americans in our study, all the following analyses were restricted to men of European descent. Five common sequence variants were genotyped in 301 patients with prostate cancer and in 250 unaffected control subjects. The case subjects were recruited from the patients who underwent prostate cancer treatment at Johns Hopkins Hospital and did not meet the criteria for HPC (N=317). The diagnosis of prostate cancer was confirmed by pathology reports, and the mean age at diagnosis was 59.3 years. Sixteen patients were excluded from the current study because of insufficient DNA. The control subjects were recruited from volunteers participating in prostate cancer–screening programs at Johns Hopkins Hospital who had a normal digital rectal examination and PSA (<4 ng/ml) (N=256). The mean age at examination was 58.5 years. Six control subjects were excluded from the current study because of insufficient DNA. All subjects in this study gave full informed consent.

The five sequence variants genotyped in this study include an SNP in the promoter region (PRO3), a 15-bp insertion/deletion of “GAATGCTTTATTGTA” in intron 1 (INDEL1), an SNP in intron 5 (IVS5-59), a missense change in exon 6 (P275A), and a 3-bp insertion/deletion of “TTA” in intron 7 (INDEL7). The positions of these sequence variants are listed in table 2. Genotyping of the three SNPs (PRO3, IVS-59, and P275A) was performed using the MassARRAY system (SEQUENOM). Genotyping of the two insertion/deletions (INDEL1 and INDEL7) was performed using the 3700 DNA Analyzer (Applied Biosystems). The primer information and detailed PCR conditions for these sequence variants are available on the authors’ Web site.

Table 2.

Frequencies of MSR1 Sequence Variants in Patients with Prostate Cancer and Unaffected Control Subjects

	No. (%) of Subjects with Genotype
SNP, (Position),aand Genotype	Control Subjects	Case Subjects
PRO3 (-14,742):
AA	214 (85.6)	235 (78.1)
AG	34 (13.6)	58 (19.3)
GG	2 (.8)	8 (2.6)
INDEL1b (-14,458):
−/−	211 (85.4	233 (78.4)
-/+	33 (13.4)	58 (19.5)
+/+	3 (1.2)	6 (2.0)
IVS5-59 (22,788):
CC	232 (92.8)	262 (87.3)
CA	18 (7.2)	34 (11.3)
AA	0 (.0)	4 (1.3)
P275A (22,850):
CC	209 (83.6)	271 (90.3)
CG	38 (15.2)	28 (9.3)
GG	3 (1.2)	1 (.3)
INDEL7 (34,504):c
−/−	204 (82.9)	264 (88.9)
−/+	41 (16.7)	33 (11.1)
+/+	1 (.4)	0 (.0)

^aPositions (bp) are based on the initiation codon (ATG) from MSR1 genomic DNA (NT_015280).

^b“+” and “−” denote with and without the 15-bp sequence “GAATGCTTTATTGTA,” respectively.

^c“+” and “−” denote with and without the 3-bp sequence “TTA,” respectively.

All five sequence variants were in Hardy-Weinberg equilibrium in control subjects, on the basis of 10,000 permutations of the Fisher probability test statistic (Weir 1996), as implemented in the software Genetic Data Analysis (GDA). The first three sequence variants (PRO3, INDEL1, and IVS5-59) were in strong linkage disequilibrium (LD), because the tests for pairwise LD among them were all highly significant (all P<.00001), again on the basis of 10,000 permutations of the exact test statistic (Weir 1996), as implemented in GDA. The last two variants also had strong LD between them (P<.00001). There was no LD between the blocks of the first three variants and the last two variants (all P>.2). The estimates of pairwise LD between all these sequence variants in case subjects and control subjects, as measured by correlation coefficients and implemented in SAS/Genetics, are presented in table 1.

Table 1.

Pairwise Linkage Disequilibrium (Correlation Coefficient) in Patients With Prostate Cancer and in Unaffected Control Subjects^[Note]

	PRO3	INDEL1	IVS5-59	P275A	INDEL7
PRO3		.99a	.59a	.03	.04
INDEL1	.98a		.58a	.03	.04
IVS-59	.69a	.71a		.08	.08
P275A	.06	.05	.06		.86a
INDEL7	.03	.02	.07	.73a

Note.— Estimates in the upper right are for control subjects and in the lower left are for case subjects.

^aIndicates P<.00001, which was based on the permutation tests of exact test statistics using GDA.

The allele frequencies of the five sequence variants were all significantly different between case subjects and control subjects, on the basis of a χ² test for allele frequencies, with 1 df (table 2; table 3). Specifically, the frequency of allele “G” of PRO3 (P=.01), 15-bp insertion (+) of INDEL1 (P=.04), allele “A” of IVS5-59 (P=.02), allele “C” of P275A (P=.01), and deletion (−) of INDEL7 (P=.04), were higher in case subjects than in control subjects, respectively. To estimate the prostate cancer risk of these variants, we performed the tests by grouping three genotypes into two genotypes as shown in table 4. Except for the INDEL7, there were significant elevated risks for prostate cancer among the sequence variants, even when adjusted for age.

Table 3.

Allele Frequencies in Patients with Prostate Cancer and Unaffected Control Subjects

	Allele Frequencies (%)
Allele	Control Subjects	Case Subjects	χ2 Test for Allele(P Values)
PRO3 “G”	7.6	12.3	.01
INDEL1 “+”a	7.9	11.8	.04
IVS5-59 “A”	3.6	7.0	.02
P275A “C”	91.2	95.0	.01
INDEL7 “−”b	91.3	94.4	.04

^a“+” denotes the presence of the 15-bp sequence “GAATGCTTTATTGTA.”

^b“−” denotes the absence of the 3-bp sequence “TTA.”

Table 4.

Odds Ratio (OR) Estimates for Prostate Cancer, Adjusting for Age^[Note]

Variant: Genotypes	OR (95% CI)
PRO3:
AA	1.00
AG/GG	1.81 (1.15–2.85)
INDEL1:
−/−	1.00
+/− or +/+	1.73 (1.10–2.72)
IVS-59:
CC	1.00
CA/AA	1.93 (1.07–3.50)
P275A:
CG/GG	1.00
CC	1.75 (1.05–2.94)
INDEL7:
−/−	1.00
+/− or +/+	1.54 (.94–2.52)

Note.— OR estimates are adjusted for age.

The haplotype analysis of these five sequence variants using the EM algorithm (Excoffier and Slatkin 1995) estimated that four major haplotypes account for >96% of all haplotypes (table 5). The haplotype frequencies were significantly different between case subjects and control subjects, with a P value of .011 from 10,000 simulations of global score tests, as implemented in haplo.score (Schaid et al. 2002). When specific haplotypes were examined, the haplotype “G (+) A C (−)” of these five variants (in the order of PRO3, INDEL1, IVS5-59, P275A, and INDEL7) had a significantly higher frequency in prostate cancer patients (6.6%) than in control subjects (2.6%), with a P value of .004 (on the basis of 10,000 simulations). It is worth noting that this haplotype did not harbor any of the rare mutations (Arg293X, His441Arg, Val113Ala, and Ile54Val) (Xu et al. 2002). For example, all eight occurrences of Arg293X and three occurrences of His441Arg resided on the haplotype “A (−) C C (−).” Therefore, it appears that the significant association between the common MSR1 sequence variants and prostate cancer risk is independent of the impact of the known rare MSR1 mutations.

Table 5.

Haplotype Frequencies of MSR1 Sequence Variants in Patients with Prostate Cancer and Unaffected Control Subjects

	Frequency (95% CI)		P values (empirical)
Haplotypea	Control Subjects	Case Subjects	Haplotype Specific
A (−) C C (−)	83.6% (80.3–86.9)	81.8% (78.7–84.9)	.44
G (+) A C (−)	2.6% (1.2–4.0)	6.6% (4.6–8.6)	.004
A (−) C G (+)	6.5% (4.3–8.6)	4.0% (2.4–5.5)	.06
G (+) C C (−)	3.7% (2.1–5.4)	4.1% (2.5–5.7)	.77
Global			.011

^aHaplotype of five SNPs (in the order of PRO3, INDEL1, IVS-59, P275A, and INDEL7)

Caution should be taken when interpreting these findings. Although the significant differences in allele and haplotype frequencies between patients with prostate cancer and unaffected control subjects could be due to the prostate cancer risk associated with these polymorphisms, it could also be due to other reasons, such as a type I error or population stratification. Regarding type I error, all the reported significance levels were nominal P values and were not adjusted for multiple comparisons. If we considered that at least 15 tests were performed in this report, and that the commonly suggested Bonferoni correction was used, none of the tests was significant at P=.05. However, the Bonferoni correction is not optimal in this case; not all of these tests were independent because of the LD between these polymorphisms and the dependence between allele and haplotype. Regarding population stratification, the results, as a case-control study, are always subject to this potential confounder: that is, the different genotype frequencies observed may partially reflect different genetic backgrounds in case subjects and control subjects. Although great attention was paid in the study design and analysis, including the restriction to subjects of European descent only, we cannot rule out the possibility of population stratification.

Each of these five sequence variants could have an important impact on MSR1 function. For example, the SNP in the promoter region and the 15-bp ins/del polymorphism could affect transcription of the MSR1 gene. The missense change of Pro275Ala could affect the function of the MSR1 protein, because it changes a conserved residue in the first Gly-X-Y repeat of the collagenous domain of the protein. However, because all five sequence variants were associated with prostate cancer risk, and the risk haplotype “G (+) A C (−)” included all the risk alleles for each variant, it is difficult to dissect genetically which variant(s) are the most important changes. It is possible that one or more of these variants are associated with prostate cancer risk. Future functional analyses using various combinations of these variants may help to confirm these findings and provide insight into the function of each variant.

The results from this study and the study of MSR1 rare mutations (Xu et al. 2002) suggest that rare mutations and common sequence variants of MSR1 confer differential risks of prostate cancer. Although the rare MSR1 mutations tend to impose relatively high risk of prostate cancer, common MSR1 sequence variants within the same major genes tend to have a relatively low risk of prostate cancer. Similar observations were observed in two other major prostate cancer susceptibility genes, HPC2/ELAC2 (MIM 605367) and RNASEL (MIM 180435) (Tavtigian et al. 2001; Carpten et al. 2002). For example, two rare mutations (E265X and M1I) of the ribonuclease L gene (RNASEL) were rare and observed mainly in families with HPC; another common sequence variant (Arg462Gln) was frequent in population and imposed intermediate risk to prostate cancer (Wang et al. 2002).

In summary, the significant differences in the allele and haplotype frequencies between patients with non-HPC and unaffected control subjects observed in this study suggest that common MSR1 sequence variants are associated with prostate cancer risk in the general population. Together with evidence that the rare MSR1 mutations are associated with increased prostate cancer susceptibility in patients with HPC and with non-HPC (Xu et al. 2002), the role of MSR1 and macrophages in prostate carcinogenesis is implicated. However, independent studies are extremely important to support these findings, given the complexity of prostate cancer.