PALB2 Variants Extend the Mutational Profile of Hungarian Patients with Breast and Ovarian Cancer.

PALB2 variants were analyzed in three independent cancer patient cohorts ( Table 1 ). Patients were referred for molecular genetic testing at our national center (Department of Molecular Genetics, Comprehensive Cancer Center, National Institute of Oncology) by clinical geneticists. The first two cohorts consisted of 1280 consecutive cancer patients with breast and/or ovarian cancer (HBOC) and 568 patients presenting with other cancers (as cancer controls), prospectively investigated between September 2021 and March 2023, regardless of their gender or age. Indication of genetic testing was established following current NCCN guideline (NCCN Clinical Practice Guidelines Genetic/Familial High-Risk Assessment: Breast, Ovarian and Pancreatic and the Hungarian Ministry of Human Resources’ professional healthcare guideline on genetic counseling (No. 20 of 2020. EüK., effective from 14 December 2020. http://www.hbcs.hu/uploads/jogszabaly/3278/fajlok/2020_EuK_20_szam_EMMI_szakmai_iranyelv_2.pdf (accessed on 1 September 2021)) [ 13 ]. Their genetic analysis was performed using a multigene panel within the routine clinical genetic care. The third, independent patient cohort was represented by 191 young breast cancer (<33 years, yBC) patients, assessed retrospectively. According to Hungarian legal and ethical regulations, germline genetic analysis was performed following genetic counseling. Each patient gave informed consent to the genetic test based on the approval of the Scientific and Research Committee of the Medical Research Council of the Ministry of Health, Hungary (ETT-TUKEB 53720-4/2019/EÜIG, ETT-TUKEB 4457/2012/EKU). To compare allele frequencies, population data from the Genome Aggregation Database (gnomAD v.2.1.1) was used applying the European non-Finnish non-cancer population (n = 134,187) (accessed on 3 July 2023) [ 14 ]. DNA extraction from peripheral blood was done by Gentra Puregene Blood Kit (#158389, Qiagen, Hilden, Germany), as previously reported [ 15 ]. Mutational profile and copy number analysis were performed using the TruSight Hereditary Cancer Panel version 2.0 (#20029551, Illumina, San Diego, CA, USA). Sequencing was run on an Illumina MiSeq instrument with MiSeq Reagent Kit v3 (600 cycles) (#MS-102-2002, Illumina, San Diego, CA, USA). All pathogenic/likely pathogenic and PALB2 variants of uncertain significance were validated on a second independently extracted DNA sample by conventional Sanger sequencing and multiplex ligation-dependent probe amplification (SALSA MLPA Probemix P260 PALB2-RAD50-RAD51C-RAD51D, MRC-Holland, Amsterdam, The Netherlands). Sanger sequencing and MLPA showed 100% concordant results with the NGS method. NGS data were analyzed by the Illumina Dragen Enrichment pipeline (v.4.0.3, San Diego, CA, USA), where both sequence variants and copy number alterations were assessed. In nucleotide detection, correct nucleotide reads with a higher than Phred quality score of 30 were accepted per position. GRCh37 genome build and MANE Select transcripts were used as reference sequences. In mapping metrics, input read number was on average 2.5 million/sample, and average % of proper reads was 95. In variant calling, variant allele frequency (VAF) between 30 and 70% was accepted for heterozygosity. Average base coverage for HBOC genes was 208 (min: 76, max: 594 reads/base). Low covered bases (<10 reads/bp) represented an average of 0.1% per gene. Variants were classified following the guidelines of the ACMG [ 16 ] and were cross-checked in the BRCA Exchange ( https://brcaexchange.org/ (accessed between 1 September 2021 and 1 June 2023)), NCBI ClinVar ( https://www.ncbi.nlm.nih.gov/clinvar/ (accessed between 1 September 2021 and 1 June 2023)), NCBI ClinGen ( https://www.clinicalgenome.org/ (accessed between 1 September 2021 and 1 June 2023)), Varsome ( https://varsome.com/ (accessed on)), and Franklin ( https://franklin.genoox.com/clinical-db/home (accessed between 1 September 2021 and 1 June 2023)) databases. Variant interpretation and cross-referencing in different databases were consecutive during patient care and were accessed between September 2021 and July 2023. Statistics were carried out by GraphPad QuickCalcs ( https://www.graphpad.com/quickcalcs/ (accessed on 12 July 2023)) and MedCalc ( https://www.medcalc.org/calc/comparison_of_proportions.php (accessed on 12 July 2023)). Depending on sample size, a two-sided Fisher exact test or an “N-1” Chi-squared test was used to compare allele frequencies between cases and population controls and to calculate 95% CIs. Age of onset curves were compared using the log-rank (Mantle–Cox) test. Results were considered statistically significant when p < 0.05.

The PALB2 protein functions as the BRCA2 partner and localizer, and it is necessary for homologous DNA recombination (HR) to repair double-strand DNA breaks. For this repair process, the BRCA1 and BRCA2 molecules must be brought together by the PALB2 protein, which serves as a link between them [ 1 , 2 ]. The formed BRCA1-PALB2-BRCA2 “tri-molecular complex” is an important component of the HR repair system that provides high-fidelity, template-dependent repair of complex DNA damages [ 3 , 4 ]. The increased risk of germline PALB2 pathogenic/likely pathogenic (P/LP) variants for breast cancer was initially suggested in 2007, and it was regarded as the third most significant breast cancer gene after BRCA1 and BRCA2 following the publication by Antoniou et al. about its breast cancer risk estimate that seemed to overlap with BRCA2 [ 5 , 6 ]. PALB2 P/LP variants represent an increased risk for female breast cancer that falls between the classic “high” and “moderate” categories. The estimated risk of P/LP PALB2 variants for female breast cancer was determined as 53% [ 6 , 7 ], while only a modestly increased risk for ovarian cancer was implied (4.8% to age 80) [ 8 ]. Interestingly, in the case of the PALB2 gene, both risk estimates are strongly influenced by family history, and the estimated absolute risk of developing cancer by age 80 years varies from 52% to 76% and 5% to 16% regarding breast and ovarian cancer, respectively, depending on the presence of familial presentation [ 3 , 9 ]. Based on these risk estimates, both the National Comprehensive Cancer Network (NCCN) and the American College of Medical Genetics and Genomics (ACMG) recommend surveillance protocols for PALB2 P/LP variant carriers, similar to BRCA1/2 P/LP variant carriers [ 3 , 9 ]. Annual mammography beginning at age 30 is endorsed, and breast MRI screening may also be considered. The role of risk-reducing mastectomy has been waiting to be determined but may be considered. Risk-reducing salpingo-oophorectomy may also be advisable in carriers at age > 45 years [ 3 , 9 ]. By these approaches, PALB2 P/LP carriers can already benefit from the advantages of primary and secondary prevention. Overall, studies suggest that patients with breast cancer harbor P/LP PALB2 variants in 0.4–3% of the cases; however, it is also suggested that the prevalence strongly varies in different populations [ 4 , 9 ]. Additionally, while some initial studies assumed associations between a PALB2 P/LP variant and increased risk of triple-negative breast cancer [ 10 , 11 , 12 ], it is now suggested that there is no established genotype–phenotype correlation, which also can be explained by the different characteristics of different populations [ 3 ]. Data on heterozygote PALB2 disease-causing variant carriers compared to BRCA1/BRCA2 carriers are still scarce in terms of both cancer incidence, spectrum of cancers and clinical outcomes [ 6 ]. The frequency of P/LP PALB2 varies, hence their clinical significance can differ among different populations [ 4 ]. Therefore, we aimed to investigate the PALB2 /LP variant prevalence and mutational spectrum in the Hungarian HBOC patients, including 191 very young cases, to compare it to patients with non-HBOC tumor types and to a healthy, non-cancer control population. We also assessed the potential effect of variants of uncertain significance and genotype–phenotype associations.

We assessed the frequency of PALB2 variants in two breast cancer cohorts and a non-HBOC cancer patient cohort ( Table 1 ). Of the 1280 consecutive HBOC patients, disease-causing variants in at least one of the breast cancer susceptibility genes were detected in 235 cases (235/1280; 18.3%). PALB2 genetic variants were identified in 58 cases (P/LP in 18 and VUS in 40 cases) ( Figure 1 a). In the 18 HBOC patients, 11 different P/LP PALB2 variants were detected ( Table 2 ). The PALB2 P/LP ratio in this cohort was 1.4% (18/1280). By comparing this finding to other breast cancer susceptibility genes recommended to be tested by the NCCN guideline [ 9 , 13 ], we found that the P/LP ratio in the PALB2 gene was the fourth most prevalent. As expected, BRCA1 was the most prevalent (68/1280; 5.3%), followed by BRCA2 (62/1280; 4.8%) and CHEK2 (26/1280; 2%). Among the consecutively referred breast cancer patients, 43 were males (vs. 1237 females). The PALB2 P/LP variant was detected only in one of them. This detection ratio was similar in male (1/43; 2.3%) and female (21/1237; 1.6%) breast cancer patients. “Double mutation” (P/LP variant in more than one HBOC susceptibility gene) was observed in 13 cases; however, none of them involved any PALB2 disease-causing genetic alteration. By analyzing genotype–phenotype associations, we found histological characteristics of the tumors of PALB2 P/LP carrier patients similar to BRCA2 carriers ( Figure 1 b). Also, age of first tumor onset and Ki67 proliferation indices in patients with the P/LP PALB2 variant were similar to BRCA2 -associated tumor patients ( Figure 1 b), and no difference in multiplex tumor occurrence among BRCA1, BRCA2 , or PALB2 carriers were observed. However, the median age of tumor onset was 44 years in PALB2 P/LP carriers versus wild-type patients where it was 48 years ( p = 0.0503, Figure 1 c) . Regarding PALB2 VUSs, we identified 40 patients carrying 24 different PALB2 VUSs ( Table 2 ). Out of the 235 positive HBOC patients, 9 carried a PALB2 variant of uncertain significance (VUS) in addition to a P/LP variant in any of the other HBOC-associated genes. To assess the potential association of PALB2 VUSs, we compared clinicopathological parameters of genetically wild-type, PALB2 P/LP, and VUS carriers. Triple-negative histology was 40% among P/LP carriers compared to 20% of the patients with wild-type and VUS carriers; however, this difference did not reach statistical significance. Regarding multiplex tumor occurrence, estrogen positivity, and HER2 positivity, no differences among the three groups with different genotypes were detected ( Figure 1 b). Among the 191 young breast cancer patients, P/LP PALB2 variants were identified in three cases (1.5%; 3/191), which did not differ from the detection ratio observed in the whole HBOC group ( Table 2 ). Incidental findings of the PALB2 P/LP variant in a patient with a non-HBOC phenotype indicate the extent of the penetrance. Therefore, we compared this in oncological and population control cohorts. In the oncological control group—patients with non-HBOC tumors—only one patient had a disease-causing PALB2 variant (0.17%; 1/568) as a secondary finding not related to the disease ( Table 2 ). This particular patient had endometriosis at the age of 34 years. The genetic testing was indicated in her case because her mother and grandmother had breast cancer at the age of 45 and 60 years, respectively; however, both of them were unavailable for genetic testing. Due to the young age of this PALB2 P/LP variant carrier, PALB2 -related tumor types could not be excluded in the future. Expectedly, HBOC patients had a higher risk of detecting PALB2 P/LP variants compared to non-HBOC patients and the non-cancer population with OR 8.1 and 9.3, respectively ( Table 3 ). We compared the PALB2 P/LP detection ratio, as an incidental finding, to the high-penetrance BRCA1 and BRCA2 genes among non-HBOC patients and found similar frequencies ( Table 4 ). While the moderate-penetrance ATM and CHEK2 disease-causing variants were more frequent as secondary findings, 0.35% and 0.44%, respectively, the difference was not statistically significant compared to PALB2 (0.088%), ( p = 0.3739 and p = 0.2175, respectively). Additionally, we assessed the frequency of PALB2 P/LP variants in the gnomAD non-cancer cohort, which showed similar detection ratios in comparison to our cancer control groups (non-HBOC phenotype) ( Table 4 ). Overall, 13 different disease-causing variants were identified in 22 patients among all the investigated cases ( Table 2 ). Interestingly, no missense genetic alteration was detected among our patients, all P/LP variants lead to loss-of-function of the PALB2 gene. The two most common variants were NM_024675.4 :c.509_510delGA (p.(Arg170IlefsTer14)) and NM_024675.4 :c.109-2A > G (p.?), which were detected in six (27%; 6/22) and three (3/22; 14%) cases, respectively ( Table 2 ). These were also reported in the NCBI ClinVar database in several cases (by 38 and 4 submitters, respectively). Of the 13 P/LP PALB2 variants, two were copy number variations affecting exons 9–10 and 11 as deletions ( Table 2 and Figure 2 ). In our cohorts, frameshift variants were more frequent compared to missense genetic alterations ( Table 5 ). No statistically significant difference was found between the distribution of the PALB2 P/LP mutation types regarding frameshift, stop, and missense variants in our samples compared to cases of the gnomAD non-cancer database, where the frameshift, stop, splice, and missense variant frequencies were 66.8%, 30.7%, 2.4%, and 0%, respectively ( Table 5 ). No correlation was identified between PALB2 variant types and the age at tumor onset, gender, single or bilateral, or multiplex disease ( Table 6 ).

We analyzed the PALB2 P/LP variant frequencies and mutational spectrum in Hungary for the first time. We investigated three cohorts of cancer patients: one as a set of consecutive patients with tumors characteristic of hereditary breast and ovarian cancer, one of young breast cancer patients (<33 years), and one cohort with non-HBOC tumors as disease controls. Additionally, we assessed the gnomAD non-cancer non-Finnish European population as a group of healthy (non-cancer) individuals. We found that the PALB2 P/LP ratio was 1.4% (18/1280) among 1280 consecutive HBOC patients. This finding was similar to Canadian, British, and Hispanic populations [ 5 , 17 , 18 , 19 , 20 ]. The overall detection rate in the literature ranged between 0.36% and 4.8% [ 4 , 18 , 21 , 22 , 23 , 24 , 25 ], and the higher indices were observed in Finland, attributed to a founder mutation [ 22 , 26 ]. Low prevalence was observed in the Jewish Ashkenazi population, in Irish, Japanese, or Dutch studies [ 27 , 28 , 29 , 30 , 31 , 32 ]. In our study cohorts, the OR for breast cancer risk in women with PALB2 P/LP variants was between 8.1 and 9.3 in breast cancer patients compared to non-HBOC cancer patients and non-cancer population, respectively. Among young breast cancer patients, we detected P/LP PALB2 variants in 1.5% (3/191), similar to studies by Cao (1.3%) and Sluiter (2%) [ 33 , 34 ]. Additionally, nearly significant differences ( p = 0.0503) were seen in the probability of the age of tumor onset between PALB2 P/LP carriers and patients with normal genotypes. Indeed, Zhou et al. found profoundly increased breast cancer risk for patients ≤ 30 years in the Chinese population compared to those > 30 years among PALB2 P/LP variant carriers [ 25 ]. “Double mutations” (P/LP variant in more than one HBOC susceptibility gene) are rare [ 35 ], and in our study, these were observed in 13 cases (1%; 13/1280); however, none of them carried any PALB2 disease-causing genetic alteration. Notably, transheterozygotes who have inherited deleterious mutations in both BRCA1 and BRCA2 were first reported in a Hungarian patient with breast/ovarian cancer [ 36 ]. Large-scale studies warrant uncovering of further transheterozygote pathogenic variants of HBOC genes and exploring the phenotype consequences of transheterozygosity. While strong genotype–phenotype associations were not found in our study, probably due to the relatively low sample numbers, bilateral breast cancer, male breast cancer, and pancreatic cancer occurred among patients carrying P/LP PALB2 variants. The more precise interrelation between these manifestations and PALB2 P/LP variants should be further investigated on a larger, independent breast cancer cohort. The prevalence of male breast cancer in our cohort was comparable (MBC/FBC: 43/1237, ratio: 0.034) to those reported in other studies (MBC/FBC: 40/2893 [ 37 ], ratio: 0.013 and MBC/FBC: 419/9675, ratio: 0.043 [ 38 ]). We found a similar PALB2 P/LP detection ratio between males (1/43; 2.3%) and females (21/1237; 1.6%). Still, for genetic counselors, it has to be considered that germline PALB2 P/LP variants were also reported in males in other studies [ 4 , 18 , 39 , 40 ] and that it represents an increased risk for developing male breast cancer (odds ratio, OR = 6.6) [ 3 , 41 ]. Also, the risk for pancreatic cancer in PALB2 heterozygote P/LP carriers is estimated to be 2–3% to age 80 years [ 3 ]. Regarding the bilateral/contralateral breast cancer risk, ACMG suggests that more systematic prospective data collection is needed to correctly address this question [ 3 ]. This year, however, a study including 15,104 prospectively followed women treated with ipsilateral surgery for invasive breast cancer reported a 35% 10-year cumulative incidence of contralateral breast cancer for PALB2 truncating variant carriers with ER-negative breast cancer [ 42 ]. Among the two control groups (cancer patients with non-HBOC tumors and healthy control, non-cancer population), we found that the PALB2 P/LP detection rate was low (0.088% and 0.076%, respectively). Also, this did not differ significantly from other HBOC susceptibility genes. This was in line with others’ findings, who also detected low frequency of PALB2 P/LP variants as incidental/secondary findings in healthy controls [ 25 , 43 ]. However, for patients with incidentally identified PALB2 variants surveillance programs according to NCCN guidelines [ 9 , 13 ] should be offered in order to early detection of potentially developing malignancies. In the Hungarian HBOC population, NM_024675.4 :c.509_510delGA and NM_024675.4 :c.109-2A > G (p.?) were detected the most frequently, in 27% (6/22) and 14% (3/22), respectively. The c.509_510delGA seems to be a common variant reported in other populations as well [ 44 ]. To determine if it can be considered as a founder variant in the Hungarian population further studies are required. Janssen et al. reported the first three exons having the highest mutation rates (exon 1 (6.3%), exon 2 (6.7%), and exon 3 (5.8%)) [ 44 ]. In our study, P/LP variants were the most frequent in exon 4 (in 5 cases, 5/22, 23%). Among our patients, frameshift, nonsense, and splice variants were the most common, while P/LP missense genetic alterations were not observed. This was similar in the healthy control cohort and in the study of Weitzel et al., who detected a similar variant distribution [ 20 ]. This may be because a significantly larger number of loss-of-function mutations (e.g., frameshift, nonsense, splice, exonic deletions/duplications) have been reported as pathogenic/likely pathogenic since the functional validation of missense variants represents a greater challenge. For this, sophisticated functional assays such as protein-protein interaction or proficiency testing in homolog recombination repair should be applied. Still, a significant number of missense variants remains unclassified; therefore, ClinGen PALB2 Variant Curation Expert Panel (VCEP) has made an effort to provide expert curation on P/LP PALB2 variants [ 3 , 4 ]. PALB2 studies have been mainly focused on truncating mutations, but the presence of variants of uncertain significance (VUS) has also been reported in patients [ 21 , 28 , 30 , 45 , 46 ], which represents a challenge for genetic counselors, clinicians, and patients as well. While we did not find differences between clinicopathological parameters of wild-type and PALB2 VUS carriers, further functional characterization of PALB2 VUSs will be able to discriminate some VUSs with pathogenic potential, hence aiding the clinical practice. Until the clarification of the role of VUSs, ACMG recommends that PALB2 VUS are not used to guide clinical management [ 3 ]. The relatively small number of PALB2 P/LP variant carriers represents a limitation in the assessment of genotype–phenotype associations and the potential additive effects of extrinsic factors such as smoking, alcohol consumption, personality type, hypertension, obesity, physical inactivity, and dietary habits on disease manifestation. Due to the low compliance of probands’ family members regarding genetic testing, the de novo rate or potential protective factors in parents cannot be assessed reliably. These should be evaluated on larger, independent breast cancer cohorts.

PALB2 P/LP variants are not rare. A total of 18 patients were identified with disease-causing variants among 1280 Hungarian HBOC patients during a one-and-a-half-year period. The c.509_510delGA variant was the most common in the studied Hungarian patient population. We did not find a significant difference in the detection ratio of PALB2 disease-causing variants according to age; however, the median age of tumor onset was lower in PALB2 P/LP carriers versus wild-type patients (44 vs. 48 years). In our cohort, the OR for breast cancer risk in women with PALB2 P/LP variants was between 8.1 and 9.3 compared to non-HBOC cancer patients and the non-cancer population, respectively. Triple-negativity was higher among P/LP carriers compared to patients with wild-type genotype and VUS carriers (40% vs. 20%); however, this did not reach statistical significance. In our patient cohort, no significant difference regarding multiplex tumor occurrence, estrogen positivity, and HER2 positivity was observed. The low rate of PALB2 incidental finding was similar to BRCA1 and BRCA2 , suggesting higher penetrance compared to ATM and CHEK2 genes. PALB2 testing is important because of the associated high cancer risk, and including patients carrying P/LP variants in preventive and screening programs can improve their life expectancy similarly to BRCA1/2 carriers.