Intraductal/ductal histology and lymphovascular invasion are associated with germline DNA-repair gene mutations in prostate cancer

Abstract

Background:

Germline mutations in genes mediating DNA repair are common in men with recurrent and advanced prostate cancer, and their presence may alter prognosis and management. We aimed to define pathological and clinical characteristics associated with germline DNA-repair gene mutations, to facilitate selection of patients for germline testing.

Methods:

We retrospectively evaluated 150 unselected patients with recurrent or metastatic prostate cancer who were offered germline genetic testing by a single oncologist using a clinical-grade assay (Color Genomics). This platform utilizes next-generation sequencing from saliva to interrogate 30 cancer-susceptibility genes. Presence or absence of a deleterious germline mutation was correlated with histological and clinical characteristics, and with family history of cancer. All patients with DNA-sequence alterations (pathogenic or variants) were offered genetic counseling.

Results:

Between July 2016 and July 2017, 150 consecutive patients underwent germline testing; pathogenic mutations were identified in 21 men (14%). Among those with germline mutations, 9 (43%) were in BRCA2, 3 (14%) were in ATM, 3 (14%) were in CHEK2, and 2 (9%) were in BRCA1. While there were no associations between germline mutations and age, tumor stage, Gleason sum or family history; mutation-positive patients had lower median PSA levels at diagnosis (5.5 vs 8.6ng/mL, P = 0.01) and unique pathologic features. Namely, men with germline mutations were more likely to harbor intraductal/ductal histology (48% vs 12%, P <0.01) and lymphovascular invasion (52% vs 14%, P < 0.01). Finally, 44% of patients with a positive germline test would not have been offered genetic screening according to current National Comprehensive Cancer Network (NCCN) guidelines.

Conclusions:

Presence of intraductal/ductal histology and lymphovascular invasion appear to be associated with pathogenic germline DNA-repair gene mutations in men with prostate cancer, and identification of these features may help to select patients for germline testing. NCCN guidelines may be inadequate in predicting which prostate cancer patients should undergo genetic screening.

1 |. INTRODUCTION

Prostate cancer (PCa) is a heterogeneous disease at the clinical, pathological, and molecular levels. Despite advances in the stratification of this disease into different risk groups (very-low risk, low-risk, intermediate-risk, and high-risk) by using PSA levels, Gleason score, and clinical stage,^1,2 the clinical course of PCa is variable. Besides high Gleason score, other pathologic characteristics are also known to affect risk and to confer an adverse prognosis, such as intraductal carcinoma (IDC)³ or ductal carcinoma⁴ histologies, lymphovascular invasion (LVI),⁵ and perineural invasion (PNI).⁶ Today, with the advent of cancer genomics, there is an ongoing effort to identify, characterize, and validate meaningful molecular biomarkers which may have a profound impact on diagnosis, therapy, and monitoring. With this better understanding, germline and somatic genetic testing may help physicians to choose whether, when and how to treat each individual prostate cancer patient in a more precise manner.

After the development of next-generation sequencing (NGS), the mutational profile of many cancers and their inherited backgrounds are now better recognized. In PCa, genomic studies have shown that the DNA-repair genes play a central role in cancer development, especially in locally advanced,⁷ and metastatic disease.^8,9 These studies demonstrate that up to 20–25% of metastatic castration-resistant prostate cancer (mCRPC) may harbor somatic DNA-repair pathway aberrations,¹⁰ and that approximately 8–12% of recurrent or advanced PCa patients may have deleterious germline abnormalities.⁹ The most common examples of these abnormal DNA-repair genes are BRCA2, ATM, CHEK2, and BRCA1.⁹ The clinical relevance of these data has increased after the observation that patients who harbor germline and/or somatic DNA-repair mutations may be more sensitive to poly ADP-ribose polymerase (PARP) inhibition.^8,11,12 Olaparib, an oral PARP inhibitor, showed clinical activity in mCRPC patients who harbored DNA-repair deficiency (DRD) mutations, both germline and somatic alterations. Even having received multiple lines of treatment, olaparib induced durable response rates including radiologic responses,⁸ underscoring the potential benefit of this class of drug in a selected (ie, DRD mutated) PCa population. Other studies have also suggested that men with germline and/or somatic DRD mutations should be managed differently from other prostate cancer patients, using unique therapeutic strategies^13,14 that may involve PARP inhibitors⁸ or PD-1 inhibitors.¹⁵

Therefore, considering the prognostic and predictive role of DRD alterations in PCa, it is important to identify clinical and pathologic characteristics that would increase the likelihood of finding such DRD mutations. By better defining which patients are “at-risk” of having a DNA-repair mutation, physicians may improve the selection of PCa patients for testing and refine the existing screening guidelines. Thus, the current study analyzed the pathologic and clinical characteristics of unselected prostate cancer patients undergoing clinical-grade germline genetic testing and aimed to correlates these characteristics with the presence or absence of deleterious germline DRD alterations.

2 |. PATIENTS AND METHODS

Unselected patients with recurrent or metastatic adenocarcinoma of the prostate who were seen as new consultations at the Johns Hopkins Hospital by a single oncologist (E.S.A.) over a one-year period formed the study population. Consecutive patients were offered germline genetic testing for clinical purposes using a CLIA-grade commercial platform (Color Genomics). Between July 15, 2016 and July 15, 2017, a total of 150 prostate cancer patients agreed to undergo germline genetic testing using this platform. Such patients spanned multiple disease states including biochemical recurrence, metastatic hormone-sensitive disease, and metastatic castrate-resistant disease. Formal genetic counseling was offered to all patients and their families if germline DNA alterations (pathogenic lesions or variants of undetermined significance [VUS]) were identified. Some patients with strong family histories of prostate or breast/ovarian cancers were also offered genetic counseling before germline testing.

This was a retrospective study that was approved by the Johns Hopkins University Institutional Review Board. Demographic, clinical, and histopathologic characteristics of all patients were collected. Family history of prostate cancer and other cancers was obtained from the medical record. The NCCN genetic/familial high-risk assessment: cancer guidelines¹⁶ and its recommendations for when to consider testing patients with localized prostate cancer for germline alterations were applied to all evaluable patients (ie, those without M1 disease at diagnosis, and with adequate family history information). We assessed the proportion of germline mutation-positive and mutation-negative patients who would have fulfilled the NCCN criteria to undergo genetic testing. The definition of a “positive” family history required at least one 1st or 2nd degree relative with a known prostate cancer or with another cancer related to DNA-repair pathways (breast, ovarian, uterine, colon, gastric, or pancreatic cancers).

2.1 |. DNA sequencing

Patients underwent next-generation sequencing (NGS) from a 30-gene panel for hereditary cancers using a saliva-based CLIA-certified commercial test offered by Color Genomics.^17,18 This platform analyzes 30 genes in which genetic alterations have been associated with an elevated risk for familial breast, ovarian, colorectal, melanoma, pancreatic, prostate, uterine, and stomach cancers. This panel included germline DNA analysis of the following genes: APC, ATM, BAP1, BARD1, BMPR1A, BRCA1, BRCA2, BRIP1, CDH1, CDK4, CDKN2A, CHEK2, EPCAM, GREM1, MITF, MLH1, MSH2, MSH6, MUTYH, NBN, PALB2, PMS2, POLD1, POLE, PTEN, RAD51C, RAD51D, SMAD4, STK11, and TP53. At least 13 of these genes are prostate cancer-susceptibility genes (BRCA2, ATM, CHEK2, BRCA1, PTEN, MLH1, MSH2, MSH6, PMS2, TP53, CDH1, RAD51C, and RAD51D).^8,9 Only pathogenic and likely pathogenic alterations according to the ClinVar database¹⁹ were considered as deleterious mutations for the purposes of this study; variants of undetermined significance (VUS) were excluded although these patients were offered genetic counseling.

2.2 |. Statistical analyses

A two-sided Fisher’s exact test was used to compare proportions for categorical variables (eg, race, tumor stage, or Gleason score) between germline mutation-positive and mutation-negative patients. An unpaired t-test was used to compare continuous variables (eg, age and PSA level) between the two groups. All statistical tests were two-sided with statistical significance set at P < 0.05. Since this study was hypothesis-generating, we did not perform Bonferroni corrections for multiple comparisons.

3 |. RESULTS

3.1 |. Patients

Between July 15, 2016 and July 15, 2017, 150 consecutive patients from the new consults clinic of one medical oncologist (E.S.A.) underwent germline genetic testing in an unselected fashion. Pathogenic inherited mutations were identified in 21 patients (14%); a list of these mutations is provided in Table 1. Variants of uncertain significance (VUS) without concurrent pathogenic alterations were identified in 30 patients (20%) (Supplementary Table S 1), while 99 patients (66%) had no deleterious germline abnormalities or VUS alleles. The prevalence of these mutations in our study as compared to the seminal paper by Pritchard et al⁹ is summarized in Table 2. Of the 21 patients testing positive for a deleterious germline mutation, the genes involved were as follows: BRCA2 (9 patients [43%]), ATM (3 patients [14%]), CHEK2 (3 patients [14%]), BRCA1 (2 patients [9%]), PALB2 (1 patient [5%]), MSH6 (1 patient [5%]), NBN (1 patient [5%]), and CDH1 (1 patient [5%]). The distribution of these pathogenic germline mutations is shown in Figure 1.

TABLE 1.

List of pathogenic^a germline mutations

Sample ID	Gene	Nucleotide change	Amino acid change	Mutation type
48	ATM	c.5932G>T	p.E1978X	Nonsense
3	ATM	c.829G>NT	p.E277X	Nonsense
25	ATM	Duplication within exon 48	-	Rearrangement
35	BRCA1	c.4357+1G>A	-	Splicing
5	BRCA1	c.4389C>A	p.Y1463X	Nonsense
2	BRCA2	c.6645delC	p.Y2215Pfs*13	Frameshift
13	BRCA2	c.7879A>T	p.I2627F	Missense
47	BRCA2	c.2808_2811delACAA	p.A938Pfs*21	Frameshift
21	BRCA2	c.4163_4164delCTinsA	p.T1388Nfs*22	Frameshift
7	BRCA2	c.462_463delAA	p.D156Rfs*2	Frameshift
11	BRCA2	c.5946delT	p.S1982Rfs*22	Frameshift
37	BRCA2	c.5946delT	p.S1982Rfs*22	Frameshift
51	BRCA2	c.5946delT	p.S1982Rfs*22	Frameshift
1	BRCA2	c.9285C>G	p.D3095E	Missense
44	CDH1	c.2195G>A	p.R732Q	Missense
33	CHEK2	c.1095+1G>T	-	Splicing
22	CHEK2	c.1427C>T	p.T476M	Missense
42	CHEK2	c.470T>C	p.I157T	Missense
24	MSH6	c.3574delG	p.V1192Lfs*3	Frameshift
31	NBN	c.657_661delACAAA	p.K219Nfs*16	Frameshift
16	PALB2	c.1050_1053delAACA	p.T351Rfs*4	Frameshift

^aDesignated as pathogenic or likely pathogenic in ClinVar.

TABLE 2.

Comparison of the prevalenve of deleterious germline mutations in this study compared to the Pritchard et al⁹ study

Gene	This study (N = 150)	Pritchard et al9 NEJM (N = 692)
BRCA2	9/150 (6.0%)	37/692 (5.3%)
ATM	3/150 (2.0%)	11/692 (1.6%)
CHEK2	3/150 (2.0%)	10/534 (1.9%)
BRCA1	2/150 (1.3%)	6/692 (0.9%)
PALB2	1/150 (0.7%)	3/692 (0.4%)
NBN	1/150 (0.7%)	2/692 (0.3%)
MSH6	1/150 (0.7%)	1/692 (0.1%)
CDH1	1/150 (0.7%)	Not analyzed
“VUS”	30/150 (20.0%)	Not reported

FIGURE 1.

Distribution of pathogenic germline mutations (N = 21)

3.2 |. Demographic characteristics

Demographic, clinical, and pathological characteristics of this patient cohort are shown in Table 3. There was no difference in median age at diagnosis between patients with and without deleterious germline mutation (61 vs 63 years, respectively, P = 0.56). Patient race (white vs non-white) was also not associated with presence or absence of a germline mutation (P = 0.48).

TABLE 3.

Baseline demographic, clinical, and pathological characteristics of our patient cohort

Patient characteristics
	Germline mutation positive	Germline mutation negative
Total no. of patients	N = 21	N = 129	P value
Median age (in years), and range
At initial diagnosis	61 (49–75)	63 (41–88)	0.56
At time of germline testing	65 (58–79)	68 (44–88)	0.22
Race, % (N)
White	80.9% (17)	89.1% (115)	0.48
Non-white	19.1% (4)	10.9% (14)
1st or 2nd degree relative, % (N)
With prostate cancer	38.1% (8)	40.3% (52)	1.00
With breast, ovarian, uterine, colon, gastric, or pancreatic cancer	52.3% (11)	51.9% (67)	1.00
Patients who fulfill NCCN criteria for genetic screening (see Table 4)
Evaluable patients (N)	18	90	0.06
Positive criteria, % (N)	55.6% (10)	20.0% (18)
Negative criteria, % (N)	44.4% (8)	80.0% (72)
Type of tissue used for histological analysis
Radical prostatectomy, % (N)	71.4% (15)	63.6% (82)	0.62
Prostate biopsies, % (N)	28.6% (6)	36.4% (47)
Clinical state at the time of germline testing
Biochemical recurrence after local therapy, % (N)	38.1% (8)	48.1% (62)	0.61
Metastatic hormone-sensitive PCa, % (N)	19.0% (4)	19.4% (25)
Metastatic castration-resistant PCa, % (N)	42.9% (9)	32.5% (42)
Tumor stage at diagnosis*, % (N)
T1/T2	33.3% (7)	34.8% (45)	0.81
T3/T4	61.9% (13)	52.7% (68)
Not reported	4.7% (1)	12.4% (16)
M1 disease at diagnosis, % (N)	14.2% (3)	23.2% (30)	0.40
Gleason sum at diagnosis, % (N)
<7	23.8% (5)	40.3% (52)	0.15
>8	76.1% (16)	58.1% (75)
Not reported	0% (0)	1.6% (2)
Presence of intraductal or ductal histology, % (N)	47.6% (10)	11.6% (15)	0.003
Presence of lymphovascular invasion, % (N)	52.3% (11)	13.9% (18)	<0.001
Presence of perineural invasion, % (N)	52.3% (11)	51.9% (67)	1.00
PSA level at diagnosis (ng/mL)
Median (range)	5.5 (1.3–22.0)	8.6 (0.9–1540)	0.01

^*Pathological tumor stage in those with prostatectomy, otherwise clinical tumor stage.

3.3 |. Family history

Neither a prostate cancer family history nor other cancers in the family were statistically associated with a positive germline test. Family history (in a 1st or 2nd degree relative) of prostate cancer was present in 38% of men (eight patients) who had a positive germline test versus 40% of men (52 patients) in the germline-negative group (P = 1.0). A family history of other DNA repair-related cancers (breast, ovarian, uterine, pancreatic, or gastric cancers) was found in 52% (11 patients) and 52% (67 patients) of germline mutation-positive and mutation-negative groups, respectively (P = 1.0).

There are no specific prostate cancer NCCN guidelines addressing germline testing. However, the NCCN genetic/familial high-risk assessment: Breastand ovarian cancer guidelines¹⁶ include recommendations for when to consider genetic testing for patients with prostate cancer. The two criteria for testing localized prostate cancer patients for BRCA½ mutations are described, as shown in Table 4. Based on these criteria, 56% of evaluable patients (with available family history, M0 disease, and Gleason ≥7 at diagnosis) with a positive germline test would have fulfilled criteria for genetic testing, while 20% of patients with a negative germline test would have also fulfilled the same criteria. Despite a trend in favor of germline-positive patients fulfilling these criteria, this association was not statistically significant (P = 0.058). More importantly, this means that 44% of men with a positive germline test would have been missed if using these NCCN guidelines to select patients for germline genetic testing.

TABLE 4.

NCCN guidelines: BRCA1/BRCA2 testing criteria

BRCA1/2 testing criteria (NCCN guidelines: BRCA-related breast and/or ovarian cancer syndromes)
• Personal history of high-grade prostate cancer (gleason score ≥7) at any age with ≥1 close blood relative with ovarian carcinoma at any age or breast cancer at ≤50 years.
• Two close blood relatives with breast, pancreatic or prostate cancer (gleason score ≥7 or metastatic) at any age.

3.4 |. Clinical and pathologic characteristics

Advanced tumor (T) stage (T3 or T4 disease) was diagnosed in the majority of men in whom the T stage was documented (133 men). Overall, 61% of patients presented with a T3 orT4 tumor. In patients who had a positive germline test, 62% (N = 13) were diagnosed with a T3 or T4 tumor; in patients who had a negative germline test, 53% (N = 68) had T3 or T4 tumors. There was no statistically significant association between T stage and the presence of a germline mutation (P = 0.81). Metastatic (M1) disease at diagnosis was also not statistically different between the mutation-positive and mutation-negative groups, respectively (14.2% vs 23.2%, P = 0.40). Overall, 91 patients (61%) had a Gleason score of >8. There was no significant association between Gleason sum and germline mutational status, despite a numerically greater number of Gleason ≥8 cases in the germline-positive group (76% vs 58%, P = 0.15). Also, there was no relationship between the presence of germline DRD mutations and Gleason score 9–10 disease when compared against Gleason ≤7 disease (data not shown).

Notably, several pathological factors that have a well-known negative impact on prognosis such as ductal/intraductal histology³ and LVI⁵ were both associated with detection of germline DRD mutations. To this end, ductal or intraductal histology was present in 48% of patients (10/21, five ductal, and five intraductal) who had a positive germline test, compared to only 12% of patients (15/129, three ductal and twelve intraductal) with a negative germline test (P = 0.003). Among those with ductal/intraductal histology, the prevalence of a germline DRD mutation was 40% (10/25); while in those without ductal/intraductal histology, only 9% (11/125) harbored germline DRD lesions.

LVI was also significantly associated with a positive germline test. More than fifty percent of patients (52%, 1½1) who tested positive in their germline test had LVI identified in their biopsy or surgical specimens, whereas LVI was only found in 14% (18/129) of those who tested negative (P = 0.0002). In patients with positive LVI, the prevalence of a germline DRD mutation was 38% (1½9); while in those without LVI, only 8% (10/121) harbored a germline DRD alteration. Another relevant prognostic factor in prostate cancer, perineural invasion (PNI),^20,21 was not associated with DNA-repair mutations. This pathologic characteristic was seen in 52% and 51% of patients with and without a deleterious germline DRD mutation, respectively (P = 1.0).

Finally, a lower PSA at initial diagnosis was associated with a positive germline test. Median PSA at diagnosis in patients who had a deleterious mutation was 5.5 ng/mL (range 1.3–22.0) versus 8.6 ng/mL (range 0.9–1540) in patients who had a germline-negative test (P = 0.012).

4 |. DISCUSSION

The findings of this study are consistent with the expected prevalence of deleterious germline DRD mutations in men with recurrent or metastatic prostate cancer, as first described by Pritchard et al.⁹ In our study, the frequency of deleterious germline mutations in this unselected prostate cancer population was 14%. These mutation rates in recurrent disease⁹ are significantly higher than the rates of BRCA½ mutations among men with localized prostate cancer.^7,22 Certainly, our analysis of a broader spectrum of DRD genes (ie, beyond BRCA1 and BRCA2) may have contributed to the higher incidence of these defects seen in our patient population, coupled with the more advanced disease states. Taking into consideration the higher risk of nodal involvement and distant metastasis of BRCA-related PCa,⁷ there may be an enrichment of germline DRD mutations in men with advanced disease. This suggests that the likelihood of harboring at least one DNA-repair gene deleterious mutation may be higher in patients with biochemical recurrence after local therapy or in those with (de novo or subsequent) metastatic disease. Strikingly, our study closely corroborates the findings of the Pritchard et al⁹ publication. The four most common germline abnormalities seen in our cohort (BRCA2, ATM, CHEK2, and BRCA1) are the same as the first four in the previously published data.⁹ Moreover, the relative distributions of the germline DRD alterations observed here are also similar to the results of the Pritchard data (Table 2 and Figure 1).

Although we did not observe a relationship between the presence of germline DNA-repair mutations and a family history of prostate or other BRCA-associated cancers, when we applied the NCCN genetic/familial breast and ovarian cancer guidelines¹⁶ for gene testing to our patients, a trend favoring the germline-positive group was noted (P = 0.058). However, despite the application of these genetic screening guidelines, 44% of men with a positive germline test would not have been offered germline testing. This suggests that the NCCN guidelines may be inadequate to capture the majority of prostate cancer patients at risk of harboring germline DRD defects, and that other clinical or pathologic criteria may better serve to identify men for genetic screening considerations.

Most interestingly, our study suggests that some readily available clinical and pathologic characteristics may be associated with the presence of deleterious DRD mutations. The presence of ductal/intraductal features or LVI, both of which are well-known adverse prognostic factors, were associated with pathogenic germline mutations. These pathological findings may serve as an early indicator to medical oncologists and urologists that germline genetic testing may be warranted. This observation also corroborates a previously published study linking intraductal histology to DRD mutations.²³ That study evaluated patient-derived xenografts of BRCA2 mutation carriers, and showed an prevalence of 42% of intraductal prostate carcinomas in BRCA2 carriers compared with a 25% prevalence in BRCA-negative patients (P = 0.002). Our current findings, coupled with this prior study, suggest that the presence of intraductal features should trigger consideration of prompt genetic counselling and germline testing for DRD mutations.

We have also shown that a lower PSA at initial diagnosis was associated with a positive germline test. Prior studies have shown that high-grade, low-PSA-producing tumors are more likely to be associated with T3-T4 disease, nodal involvement and distant metastases at the time of diagnosis,²⁴ similar to BRCA½-positive prostate cancers.⁷ We speculate that patients with high-grade disease which manifests with low PSA levels may be more likely to harbor DNA-repair gene mutations. In addition, some (but not all) studies have reported a non-statistically significant trend between intraductal/ductal prostate cancer histologies and lower median PSA levels.^25,26 Further study is needed to confirm these preliminary findings.

Our study has several limitations that should be considered when interpreting the results. First, this was a retrospective study. While several provocative associations have been demonstrated, causal relationships are difficult to prove. Second, other inherent limitations of this type of study are the introduction of selection bias and information bias. We tried to mitigate selection bias by including consecutive patients from the outpatient clinic of one medical oncologist who did not recommend germline testing based on family history of other clinical information. Third, and specifically relevant to the associations between genotype and histologic features, pathologic materials were not centrally reviewed by a single pathologist. We therefore relied on the clinical interpretations of a range of anatomic pathologists at the Johns Hopkins Hospital. However, this does increase the real-world value of these results if the diagnosis of intraductal/ductal features and PNI can be made reliably by a wide range of pathologists. Finally, this study grouped together ductal and intraductal cases for statistical analysis, and most pathologists would consider these to be separate histological entities.

5 |. CONCLUSION

This study reports positive associations between the presence of germline DRD mutations and several clinicopathologic features: ductal/intraductal histology, lymphovascular invasion, and a low PSA level at initial presentation. These factors could influence medical decisions concerning the recommendation for germline genetic testing with a focus on inherited DNA-repair gene alterations. The confirmation of such DRD mutations would have significant implications both for family counseling as well as therapeutic choices for these patients. We believe that the presence of one or more of these factors could be added to the family history when considering the decision to request genetic testing and counseling. If these data are confirmed, we would propose that all prostate cancer patients with intraductal/ductal histologies should consider germline genetic testing for DRD mutations. Alternatively, such patients could undergo tumor-based somatic DNA sequencing, followed by germline DNA sequencing in those with a tumoral DRD alteration.