DNA Damage Repair Defects and Survival Outcomes for Patients With Resected Pancreatic Ductal Adenocarcinoma

To the Editor:

Whole-genome analyses identified DNA damage repair (DDR) alterations in up to 24% of pancreatic ductal adenocarcinomas.^1,2 Genetic defects in BRCA1/2 and other key homologous recombination repair (HRR) genes such as PALB2, RAD51C/D, ATM, and ATR may cause homologous recombination deficiency (HRD).³ Homologous recombination deficiency has been predictive of platinum response in ovarian, fallopian, and peritoneal carcinomas, and correlated with improved survival after platinum chemotherapy.⁴ The prognostic value of HRD in pancreatic cancer has not been thoroughly investigated, but mutations in several HRR genes such as BRCA1/2, ATM, ATR, CHEK2, PALB2 may predict benefit from platinum-based chemotherapy.^5,6 In this study we evaluated the prevalence and prognostic value of DDR defects in patients with resected pancreatic cancer.

METHODS AND MATERIALS

This retrospective cohort study included 100 consecutive patients who underwent surgical resections for pancreatic cancer at the University of Washington (UW) Medical Center between 1999-2008. Tumor samples were obtained from UW Northwest BioTrust Biorepository and were required to contain at least 10% tumor tissue. Tumor DNA was sequenced using BROCA-HR,⁷ and BRCA1 and RAD51C methylation was assessed by methylation sensitive PCR as previously described.⁸

DNA damage repair defects were categorized in the following groups based on the degree of HR deficiency: 1) BRCA1/2 core HRR; 2) non-BRCA1/2 core HRR: PALB2, ATM, ATR, RAD51C/D, BRIP1, BARD1; 3) non-core HRR: CHEK1/2, ERCC1/4, FANCA-M, RAD51, RECQL; 4) other DDR; 5) HRR wild type (includes other DDR genes defects); 6) DDR wild type. The core HRR group included BRCA1/2 and non-BRCA1/2 genes, and the group with any HRR defects included core and non-core HRR genes. Homologous recombination and DNA damage repair status was considered wild type if mutated gene variant allele frequency (VAF) was less than 5%. Survival was calculated from diagnosis to date of death and patients last known to be alive were censored at last follow-up. The log-rank test was used to compare survival between DDR groups of interest.

RESULTS

A total of 81 (81%) tumor samples had histologically confirmed pancreatic cancer with adequate DNA for sequencing and were included in the analysis. Median age at diagnosis was 62 years (range 38-85). No patients met clinical criteria for familial pancreatic cancer or reported Ashkenazi Jewish ancestry. Most patients (n = 80) had stage 1/2 disease per American Joint Committee on Cancer (AJCC 7^th edition). Twelve patients (15%) were known to have received platinum chemotherapy (platinum exposed) during their lifetime.

Twenty-three (28%) pancreatic cancers harbored DDR gene defects. Eight (10%) tumors had core HRR defects: six deleterious BRCA1/2, of which three were sequences previously identified as pathogenic germline mutations; one each PALB2, ATM (with concurrent gBRCA2 defect), and ATR. Eight (10%) non-core HRR gene defects were identified: CHEK2 (n = 2), ERCC4 (n = 2; one concurrent BRCA1 defect), FANC-A, FANC-B, FANC-M, and RECQL (n = 1 each). Fourteen (17%) other DDR gene defects (four concurrent with BRCA1/2 or ATR mutations) occurred among 12 tumors: KMT2C/MLL3 (n = 4), KMT2D/MLL2 (n = 2), POLQ (n = 2), and CDH4, ERCC6, MSH6, PRKDC, RINT1, XPC (n = 1 each).

Eighty patients were included in the survival analysis, as one patient died from postoperative complications within 30 days from surgery and was excluded. Characteristics and survival data for 23 patients with DDR gene alterations are shown in Table 1. The median survival was 23.2 months (95% confidence interval [CI], 18.4–26) for all patients (n = 80), and individual groups had median survival of 37.1 months (95% CI, 5.0–146.7), 37.5 months (95% CI, 5.9–not reached), 22.2 months (95% CI, 6.5-55.5), 23.2 months (95% CI, 18.1–25.8), and 23.2 months (95% CI, 17.6–26.9) for patients with BRCA1/2 defects (n = 6), non-core HRR defects (n = 7), other DDR defects (n = 8), HRR wild-type (n=65), and DDR wild-type (n = 57), respectively. No significant survival differences were observed among patients with core (P = 0.94) or any HRR gene defects (P = 0.53) versus HRR wild-type tumors (Fig. 1).

TABLE 1.

Characteristics and Overall Survival for 23 Patients With DDR Gene Defects

Pt	HRR Category	HRR Gene	Defect	Other DDR Gene	Defect	Family History	Chemo	Age , y	OS, mo
1	BRCA1/2	BRCA *	c.68_69delAG	POLQ *	c.4289T>A p.L1430X	None	N/A	77	5
2	BRCA1/2	BRCA1	c.3600G>C	XPC	c.2083dupC	Leukemia†	Gem + Kras vaccine,‡ FFOX,§ GnP§	57	86.2
3	BRCA1/2	BRCA * ATM ∥	c.1A>G c.4110-1delG			Breast¶ ovarian¶	N/A	63	11.9
4	BRCA1/2	BRCA1 ERCC4	c.541G>T c.2405G>A	CHD4	c.2764G>T	Prostate¶ Lung¶	GTX‡§# GemOx‡ XE§	45	59.6
5	BRCA1/2	BRCA1	Methylation			None	N/A	65	146.7
6	BRCA1/2	BRCA2 *	c.2806_2809delAAAC			Ovarian†	Yes,# N/A	43	14.5
7	Non-BRCA1/2 core HRR	ATR	c.5347C>T	KMT2C KMT2D	c.6654C>G c.16605G>A	H&N¶	N/A	72	19
8	Non-BRCA1/2 core HRR	PALB2	c.2815_2822delTTGGAAAT			None	None	85	6.8
9	Non-core HRR	CHEK2 *	c.538C>T			Lung†	Yes,‡ N/A	60	49
10	Non-core HRR	CHEK2 *	c.190G>A			Renal† Gastric†	None	78	55.2
11	Non-core HRR	ERCC4	c.148C>T			Colon,† Lung,† Liver,† H&N†	GTX,‡ GemOx,§ XE§	69	26
12	Non-core HRR	FANCA	c.4171+1G>A			None	Yes,‡ N/A	65	19.3
13	Non-core HRR	FANCB	c.1327-2insT	RAD51D	c.121C>T (low VAF)	None	GX‡	52	124**
14	Non-core HRR	FANCM	c.5926_5927delTA			Pancreas†	N/A	59	1.8**
15	Non-core HRR	RECQL	c.643C>T			None	N/A	42	5.9
16	Other DDR			ERCC6	c.3939delC	None	Yes,‡ N/A	38	6.5
17	Other DDR			KMT2C	c.2646delC	None	N/A	70	55.4
18	Other DDR			KMT2C PRKDC	c.3274C>T c.5079C>G	None	GemOx,# Gem‡	51	13.7
19	Other DDR			KMT2C	c.8390dupA	Breast† Pancreas¶	Yes,‡ N/A	58	34.8**
20	Other DDR			KMT2D	c.5577_5584delCCCTGAGA	Breast†	N/A	83	25.2
21	Other DDR			MSH6	c.3254dupC	None	Gem + Kras vaccine‡	51	16.4
22	Other DDR			POLQ *	c.5857C>T	Pancreas†	GTX,§ Gem-Cis,§ Gem-iri,§ CAPOX§	52	19.2
23	Other DDR			RINT1	c.1659G>A	Prostate† Brain¶	5FU‡	57	25.2

^*Known pathogenic germline mutation

^†First degree family member

^‡Adjuvant

^§Palliative

^∥Germline

^¶Second degree family member

^#Neoadjuvant

^**Survival duration from the date of diagnosis to last follow up (no date of death available).

CAPOX, capecitabine + oxaliplatin; Cis, cisplatin; DDR, DNA damage repair; E, erlotinib; 5FU, 5-fluorouracil; FFOX, FOLFIRINOX; FHx, family history; Gem, gemcitabine; GnP, gemcitabine/nab-paclitaxel; GTX, gemcitabine + docetaxel (taxotere) + capecitabine (xeloda); GX, gemcitabine + capecitabine (xeloda); H&N, head and neck; HRR, homologous recombination repair; Iri, irinotecan; N/A, not available; Ox, oxaliplatin; Pt, patient; VAF, variant allele frequency (low <5%); X, capecitabine (xeloda)

FIGURE 1.

Kaplan Meier curves of overall survival by HRR defects. HRR, homologous recombination repair; WT, wild type; core HRR, BRCA1/2 + non-BRCA1/2 genes (PALB2, ATM, ATR); Any HRR, core (BRCA1/2/PALB2/ATM/ATR) + non-core HRR genes (CHEK2, ERCC4, FANC-A, FANC-B, FANC-M, RECQL); WT HRR, no HRR gene defects.

DISCUSSION

In this retrospective cohort of patients with resected pancreatic cancers, DDR genetic defects were identified among 28% of cancers, including 19% with HRD. Patients carrying any homologous recombination repair defects, most platinum-naïve, had comparable survival with those with wild-type tumors (22.7 vs 23 months). Similarly, Golan et al noted no survival differences for gBRCA carriers compared to sporadic tumors,⁹ while Pishvaian et al reported longer survival only for platinum-exposed pancreatic cancers harboring HRR defects, suggesting that HRD has similar clinical implications with BRCA1/2 defects.⁵ Although limited by a small sample size, our analysis did not demonstrate a prognostic effect from BRCA1/2/PALB2 or other HR-DNA damage repair gene defects for resected PDA patients. Pancreatic cancer patients with HR-DDR genomic alterations should be included in clinical trials with platinum chemotherapy, and/or molecularly targeted therapies such as poly (ADP-ribose) polymerase, ATR, and other DDR-inhibitors, given the urgent need to optimize treatment outcomes.