|Home | About | Journals | Submit | Contact Us | Français|
Germline mutations in the PALB2 gene are associated with an increased risk of developing breast but little is known about the frequencies of rare variants in PALB2 and the nature of the variants that influence risk. We selected participants recruited to the Women’s Environment, Cancer, and Radiation Epidemiology (WECARE) Study and screened lymphocyte DNA from cases with contralateral breast cancer (n = 559) and matched controls with unilateral breast cancer (n = 565) for PALB2 mutations. Five pathogenic PALB2 mutations were identified among the cases (0.9%) versus none among the controls (p=0.04). The first degree female relatives of these five carriers demonstrated significantly higher incidence of breast cancer than relatives of non-carrier cases, indicating that pathogenic PALB2 mutations confer an estimated 5.3 fold increase in risk (95% CI: 1.8–13.2). The frequency of rare (<1% MAF) missense mutations was similar in both groups (23 versus 21). Our findings confirm in a population-based study setting of women with breast cancer the strong risk associated with truncating mutations in PALB2 that has been reported in family studies. Conversely, there is no evidence from this study that rare PALB2 missense mutations strongly influence breast cancer risk.
PALB2, originally identified as a BRCA2-interacting protein, is crucial for key BRCA2 genome caretaker functions (Xia, et al., 2006) and more recently has also been shown to interact with BRCA1(Sy, et al., 2009a; Sy, et al., 2009b; Zhang, et al., 2009). Germline mutations in PALB2 (MIM# 610355), are associated with increased risk of breast (Rahman, et al., 2007) and mutations have been identified in approximately 1% of hereditary breast cancer (HBC) families throughout the world, summarized in (Tischkowitz and Xia, 2010). The first study showing an association with germline mutations in PALB2 and breast cancer estimated the relative risk to be 2.3 (95% confidence interval (C.I.) 1.4–3.9) (Rahman, et al., 2007), but subsequent studies have suggested that the risk could be higher (Erkko, et al., 2008; Southey, et al., 2010). In this investigation we examine the breast cancer risk conferred by PALB2 mutations in a population-based case-control study of high risk women, the WECARE Study (Bernstein, et al., 2004). In this study, mutation frequencies in incident cases of contralateral breast cancer (CBC) are compared with the population from which CBC cases emerge, namely survivors of unilateral breast cancer (UBC). This study design is especially advantageous for studying rare genetic mutations, since the frequencies of true risk variants are higher in both cases and controls as both have breast cancer (Begg and Berwick, 1997). Other HBC genes such as BRCA1/BRCA2 and the CHEK2*1100delC variant have been studied in this way and have demonstrated increased risk of contralateral breast cancer (CBC) (Fletcher, et al., 2009; Malone, et al., 2010). We screened the coding regions and adjacent intronic areas of the PALB2 gene in lymphocyte DNA from BRCA1/BRCA2-negative women participating in the WECARE study to determine the mutation frequency in CBC cases compared to UBC controls.
The Women’s Environment, Cancer, and Radiation Epidemiology Study is a population-based, nested case-control study of CBC, which has been described previously (Bernstein, et al., 2004). All WECARE Study participants were ascertained through one of five population-based cancer registries covering the country of Denmark along with the State of Iowa, Los Angeles County and the Orange County-San Diego regions of California, and three western Washington counties in the United States (the US registries participate in the Surveillance, Epidemiology, and End Results [SEER] registry system). All participants were diagnosed before age 55 years between 1985 and 2000 with a first primary invasive breast cancer that had not spread beyond regional lymph nodes. Cases were diagnosed with a second primary invasive or in situ CBC at least one year after first primary diagnosis in the period 1986 to 2001; controls were women with unilateral breast cancer (UBC) who did not develop a second primary breast cancer during the study period (until 2001). For this investigation we selected cases and controls that had been previously screened for pathogenic BRCA1 and BRCA2 mutations by denaturing high-performance liquid chromatography (Malone, et al., 2010) and excluded individuals with detectable pathogenic mutations in these genes according to the following criteria: 1) exon sequence changes predicted to truncate protein production; 2) splice site mutations located within two base pairs classified according to the of an intron/exon boundary or shown to cause aberrant splicing; 3) missense changes with known pathogenic functional effects. Our final analytic group consisted of 559 CBC cases and 565 UBC matched controls and the characteristics of the study participants are given in Table 1.
Genomic DNA was obtained from lymphocytes using standard extraction methods. The 13 coding exons of PALB2 (NCBI reference sequence NM_024675.3) were screened using high-resolution melt analysis (HRM), a method with similar sensitivity to sequencing (Wittwer, 2009). All work was carried out in a single laboratory using the conditions listed in Table 2. The PCR for HRM was performed in 5ul total volume (2ul mastermix, 0.5ul LCGreen, 0.5ul each of forward and reverse primers (2.5uM stock concentration), 0.5ul of double distilled sterile water, 1ul of DNA (25ng/ul). As indicated in Table 2, Q solution instead of water was used for the exon 1 PCR and 25mM of MgCl2 was added to the PCRs for exons 5.1, 13.1 and 13.2. The PCR protocol was as follows (for 66°C annealing temperature): 95°C for 2min, 95°C for 45sec, 73°C 45sec Touchdown (-1°C increment for 7 cycles), followed by 95°C for 45sec, 66°C for 45sec for 35 cycles. The PCR products were then analysed by HRM using the LightScanner instrument (Idaho Technologies, Salt Lake City, Utah) according to the manufacturer’s protocol. Nucleotide numbering reflects cDNA numbering where +1 corresponds to the A of the ATG translation initiation codon in the reference sequence (NM_024675.3) and the initiation codon is codon 1. Variants were confirmed by Sanger sequencing and were classified as pathogenic using the same criteria as the BRCA1/BRCA2 study (Malone, et al., 2010) mentioned above. Sequence changes that did not fulfill these criteria for being pathogenic were divided into three categories 1) missense variants, 2) exonic synonymous single nucleotide variants (“synonymous”), and 3) intronic variants. We further examined the effects of missense mutations, classified on the basis of bioinformatic tools that predict functional significance: Sorting Intolerant from Tolerant (SIFT) (Kumar, et al., 2009), Polyphen (Adzhubei, et al., 2010) and Align GVGD (Tavtigian, et al., 2008). SIFT and Polyphen were used under default conditions and with the programs’ internally generated alignments. Align-GVGD was used with a curated alignment in which the sequence furthest diverged from human was from the fish Danio rerio. This alignment, or updated versions thereof, is available at the Align-GVGD website http://agvgd.iarc.fr/alignments.
First, we evaluated the broad case-control associations with respect to various, possibly overlapping mutation types using individuals who had wild-type PALB2 as the baseline and estimating the risk of carrying a deleterious, missense synonymous, or intronic mutation. Due to the small sample sizes in some of the analyses presented, we employed Fishers exact test and exact (i.e. permutation-based) methods for constructing confidence intervals using Cytel Studio© software (Cytel Inc., Cambridge, MA, USA). The exact two-sided Cochran-Armitage trend test was employed to investigate the relationship between the case-control status and the bioinformatics tools classifications and the exact Kruskal-Wallis test was performed to test for differences in continuous patient characteristics by mutation type. We estimated the relative risk conferred by pathogenic PALB2 mutations by comparing the breast cancer incidence in their first degree female relatives with the corresponding incidence in first degree relatives of non-carrier CBC cases using conventional actuarial techniques, and then by transforming the resulting estimate (Saunders and Begg, 2003). We further examined the relative risks of individual rare missense variants (MAF≤1%) using a hierarchical modeling approach developed for this purpose (Capanu, et al., 2011). This model included adjustment for eight common polymorphisms (MAF>1%), the presence of a truncation or splicing mutation, the presence of a rare intronic mutation, the presence of a rare synonymous mutation, as well as childbearing history as possible confounders. The effects of individual variants were adjusted on the basis of the average age at diagnosis of the carriers of the specific variant, family history of breast cancer in first- degree relatives for these carrier probands, as well as the three bioinformatic prediction tools.
We screened the PALB2 gene in a total of 1124 women with breast cancer and we identified five pathogenic mutations in the cohort of 559 women with CBC versus 0/565 women with UBC (p=0.04), Table 3. Of the five mutations, three have been previously reported: 1592delT is a known founder mutation in Finland (Erkko, et al., 2007), while W1038X and Y1183X have been reported by several groups (Table 4). Among the five carriers of these mutations, the median ages of the first and second breast cancers were 46 and 55 years respectively and all probands had at least one first-degree relative affected with breast cancer, although none had an additional young-onset case (under age 50 years) in the family. Nonetheless the breast cancer incidence in these 17 first degree female relatives of PALB2 carriers was significantly higher than the corresponding incidence in first degree relatives of non-carrier cases, and from this we derived a relative risk estimate for carriers of a pathogenic mutation of 5.3 (95% CI: 1.8–13.2). The majority of the breast cancers in these five women were ER positive, PR positive infiltrating ductal type, with the notable exception of the carrier for the 1592delT mutation where both tumors were ER negative, PR negative, the first being classified as medullary-type (Table 4).
We found no statistically significant difference in the frequency of missense variants between cases and controls (these variants are listed in Supp. Table S1). However, when classified by their predicted degrees of pathogenicity using the three different models, SIFT (Kumar, et al., 2009), Polyphen (Adzhubei, et al., 2010) and Align GVGD (Tavtigian, et al., 2008), for which the criteria was set as GV=0 and GD≥65, following the classification at http://agvgd.iarc.fr/classifiers.php), there were marginally significant trends for SIFT and align-GVGD, evidenced by the higher relative frequencies of cases in the highest risk categories (Table 3). Further analyses of the individual rare missense mutations using hierarchical modeling revealed no specific mutations as significantly associated with risk (data not shown).
The overall frequency of intronic variants was higher in cases than controls (Table 3), and we did observe a notably increased frequency of rare intronic variants (MAF<0.1%) with 12 rare intronic variants seen in cases versus one variant in controls (RR=13.7, 95% CI 2.0, 591.3) listed in Supp. Table S2. This finding remained significant even when rare and common intronic variants were combined (p=0.01). Further in silico analysis of these intronic variants using Human Splicing Finder v2.4.1 (Desmet, et al., 2009) did not predict that they would have significant effects on splicing. Furthermore, when we examined the incidence of breast cancer in first degree relatives of carriers of rare intronic variants the rate was not elevated compared to non-carriers. Eleven of the 12 individuals were of non-Hispanic white ethnicity making an ethnic-specific effect unlikely. When the median age of onset of the second breast cancer was compared between the rare intronic group (51.8 years, n=11), carriers of deleterious mutations (56.6 years, n=5) and the rest of the cases (age 51.1, n=543) no significant difference was observed (p=0.3, Kruskal-Wallis test). There was no significant effect of having received radiation therapy among subsets of mutation carriers (number of events was too small to analyze deleterious variants independently; data not shown).
A number of studies have implicated PALB2 gene mutations as a rare, but important, contributing factor to hereditary breast cancer (Tischkowitz and Xia, 2010). The results presented here support these findings in a population-based setting, and confirm that women with germline PALB2 mutations also have an increased risk of CBC. Although only five clearly pathogenic mutations were identified in the group of 1124 women, all of these occurred among the cases. The penetrance of PALB2 gene mutations was originally estimated to be associated with a 2.3-fold increased risk (Rahman, et al., 2007) but subsequent studies suggested this could be higher, at least for specific mutations (Erkko, et al., 2008; Southey, et al., 2010). The relative risk estimate of 5.3 from our study is indeed higher, although this estimate needs to be interpreted with caution given the small number of carriers and carrier relatives in our study and the correspondingly wide confidence intervals.
Despite a large number of published studies, to date no definitely pathogenic PALB2 missense variants have been identified, suggesting that they are non-existent or rare. Our results are consistent with this literature. As an analogy, in the breast cancer predisposing genes BRCA1 and BRCA2 the vast majority of missense variants are neutral, but a few deleterious variants have been identified. The identification of isolated, individual deleterious mutations using case-control studies requires very large numbers of subjects, or evidence that particular types of missense variants are deleterious collectively, allowing aggregation of the data to create sufficient statistical power. Our hierarchical modeling (data not shown) was designed for this purpose, but did not uncover any trends of this nature. Thus, our study does not rule out the possibility that isolated missense PALB2 mutations may be deleterious, but it offers no strong evidence that this is a likely possibility.
The clinical significance of our observation that rare intronic variants are more frequent in the CBC cohort is unclear as none of these variants were predicted to be pathogenic by affecting splicing and only one has been reported previously. Further investigations to address this would involve studying these variants in additional breast cancer cohorts and, at the cellular level, looking for possible functional effects such as diminished BRCA2 binding capacity and reduced homologous recombination efficiency (Tischkowitz, et al., 2007). It should also be noted that we did not see an increased incidence of breast cancer in first degree relatives of carriers of rare intronic variants which would argue against these being pathogenic.
Our study was conducted using population-based series of breast cancers and as such it avoids the potential problem of ascertainment bias that can occur in cohorts derived from familial cancer clinics or other high-risk settings. Nevertheless, our study has some limitations. First, while a large number of women were screened for PALB2 mutations, the relative rarity of these mutations meant that only a small number of pathogenic mutations were identified. Second, because the study excluded women with synchronous cancers and women who underwent prophylactic contralateral mastectomy following initial breast cancer, and included only women who survived their initial breast cancer, the results may underestimate the overall mutation prevalence. Lastly, we used High Resolution Melting analysis which, as with conventional sequencing, would miss detection of large exon deletions. However, based on our previous experience on the rarity of PALB2 exon deletions (Tischkowitz, et al., 2009) it is unlikely that we would have missed a significant number of mutations.
In summary, we have confirmed that germline deleterious truncating PALB2 mutations are associated with breast cancer in a population-based case-control study. Although PALB2 mutations are a rare cause of breast cancer, their overrepresentation in the cohort of women with CBC is relevant to the clinical management of newly diagnosed women with breast cancer who are found to be PALB2 mutation carriers as it implies a significant risk of developing a second breast cancer. Our study also provides suggestive evidence that rare intronic variants may be associated with risk, but this finding requires careful validation.
This work was funded by the Jewish General Hospital Weekend to End Breast Cancer, the Quebec Ministry of Economic Development, Innovation and Export Trade and the Jodi Taiger Lazarus Fund for Hereditary Breast Cancer Research. MT holds a Fonds de la Recherche en Santé du Québec (FRSQ) clinician-scientist award. Additional support: National Cancer Institute, Grant numbers CA131010, R01 CA097397, U01 CA083178. The WECARE Study Collaborative Group - Memorial Sloan Kettering Cancer Center (New York, NY): Jonine L. Bernstein Ph.D. (WECARE Study P.I.), Colin Begg. Ph.D., Marinela Capanu Ph.D., Xiaolin Liang M.D., Anne S. Reiner M.P.H., Irene Orlow Ph.D; City of Hope (Duarte, CA) Leslie Bernstein Ph.D. (sub-contract P.I.), Laura Donnelly-Allen (some work performed at University of Southern California, Los Angeles CA); Danish Cancer Society (Copenhagen, Denmark): Jørgen H. Olsen M.D. DMSc. (Sub-contract P.I.), Lene Mellemkjær Ph.D.; Fred Hutchinson Cancer Research Center (Seattle, WA): Kathleen E. Malone Ph.D. (Sub-contract P.I.), International Epidemiology Institute (Rockville, MD) and Vanderbilt University (Nashville, TN): John D. Boice Jr. Sc.D. (Sub-contract P.I.); Lund University (Lund, Sweden): Åke Borg Ph.D. (Sub-contract P.I.),Theresa Sandberg Ph.D.; National Cancer Institute (Bethesda, MD): Daniela Seminara Ph.D. M.P.H; New York University (New York, NY): Roy E. Shore Ph.D., Dr.P.H. (Sub-contract P.I.); Northern California Cancer Center (Fremont, CA): Esther John Ph.D. (Sub-contract PI; Norwegian Radium Hospital (Oslo, Norway): Anne-Lise Børresen-Dale Ph.D. (Sub-contract P.I.); Samuel Lunenfeld Research Institute, MSH (Toronto, Canada): Julia Knight, Ph.D. (Sub-contract P.I.), Anna Chiarelli Ph.D. (Co-Investigator); Translational Genomics Research Institute (T-Gen)(Phoenix, AZ): David Duggan Ph.D. (Sub-contract P.I.); University of California at Irvine (Irvine, CA): Hoda Anton-Culver Ph.D. (Sub-contract P.I.); University of Iowa (Iowa City, IA): Charles F. Lynch M.D., Ph.D. (Sub-contract P.I.); University of Southern California (Los Angeles, CA): Robert W. Haile Dr.P.H. (Sub-contract P.I.), Daniel Stram Ph.D.(Co-Investigator), Duncan C. Thomas Ph.D. (Co-Investigator), Anh T. Diep (Co-Investigator), Shanyan Xue M.D., Nianmin Zhou, M.D, Evgenia Ter-Karapetova; University of Texas, M.D. Anderson Cancer Center (Houston, TX): Marilyn Stovall Ph.D. (Sub-contract P.I.), Susan Smith M.P.H. (Co-Investigator);University of Virginia (Charlottesville, VA): Patrick Concannon, Ph.D. (Sub-contract P.I.), Sharon Teraoka, Ph.D. (Co-Investigator).
Supporting Information for this preprint is available from the Human Mutation editorial office upon request (humu/at/wiley.com)