mRNA Expression and Methylation of the RAD51,ATM,ATR,BRCA1,and BRCA2 Genes in Gastric Adenocarcinoma

Abstract

Background:

Immunohistochemical prognostic significance of the homologous recombination-related proteins RAD51, ATM, BRCA1, and BRCA2 is known in gastric adenocarcinoma, one of the deadliest cancers.

Objective and design:

This retrospective cohort study aimed to evaluate mRNA expression and promoter methylation of some homologous recombination-related genes in this neoplasm.

Methods:

We evaluated mRNA expression and methylation of RAD51, ATM, ATR, BRCA1, and BRCA2 in tumor and non-tumor frozen samples from gastrectomy specimens by RT-qPCR and MS-HRM, correlating our results with previous immunohistochemistry data and prognostic features.

Results:

RAD51, ATR, BRCA1, BRCA2, and ATM mRNA expression was detected in 93.75% (45/48), 93.75% (45/48), 91.67% (44/48), 83.33% (40/48), and 89.58% (43/48) of the tumors; partial or complete methylation, in 94.87% (37/39), 0 (0/42), 97.56% (40/41), 100% (41/41), and 0 (0/40), respectively. Most gene pairs showed significant weak to moderate positive correlations of tumoral mRNA expression with each other: RAD51 with ATR (P = .027), BRCA1 (P < .001), and BRCA2 (P < .001); ATR with BRCA1 (P = .007), and ATM (P = .001); BRCA1 with BRCA2 (P = 0.001). BRCA1 mRNA was reduced in tumors compared with non-neoplastic mucosa (0.345 vs 1.272, P = .015) and, excluding neoadjuvant therapy cases, in T3 to T4 tumors compared with T2 (0.414 vs 0.954, P = .035). Greater tumoral RAD51 mRNA levels correlated with perineural invasion (1.822 vs 0.725, P = .010) and death (1.664 vs 0.929, P = .036), but not with survival time. There was an inverse association between nuclear immunohistochemical positivity for ATR and its mRNA levels (0.487 vs 0.907, P = .032), and no significant correlation for the other markers.

Conclusions:

Our results suggest RAD51, BRCA1, and BRCA2 methylation as a frequent epigenetic mechanism in gastric cancer, support the hypothesis that reduced BRCA1 expression participates in disease progression, and show an association between RAD51 mRNA and perineural invasion and mortality that may be considered unexpected, considering the former immunohistochemical studies. The lack of correlation between immunohistochemistry and mRNA, and even the inverse association, for ATR, can be seen as indicative of action of post-transcriptional or post-translational regulatory mechanisms, to be better investigated.

Keywords

Stomach neoplasms homologous recombination DNA repair mRNA methylation

Introduction

Malignant tumors of the stomach are the fourth major cause of death due to cancer worldwide.¹ More than 90% of them are adenocarcinomas.² The pathogenesis of gastric adenocarcinoma is complex, with infection of the gastric mucosa by Helicobacter pylori being the main risk factor. The mechanisms by which it induces carcinogenesis are several and include an increased production of reactive species caused by inflammation, which leads to DNA damage,³ including double-strand breaks.⁴

The most accurate repair mechanism for DNA double-strand breaks is homologous recombination. ATM, BRCA1, BRCA2, ATR, and RAD51 proteins participate in this repair pathway. Initially, DNA breakage activates ATM kinase. Then, this kinase phosphorylates BRCA1, which acts on the resection of the 5′ ends of the break, a process that activates and is stimulated by ATR via positive feedback.⁵ RAD51 covers the newly formed single-stranded ends, in a BRCA2-dependent process, and participates in the search for the homologous sequence in the sister chromatid that serves as a template for DNA repair.⁶

Mutations in genes encoding these proteins are uncommon in gastric adenocarcinoma.⁷ However, changes in their expression are frequent and correlate with clinicopathological characteristics, as shown by several previous studies, mostly based on immunohistochemistry.^8-18

In gastric cancer, the relationship between the methylation of several genes and pathogenesis, prognosis, diagnosis, and resistance to chemotherapy has been reported.¹⁹ However, the methylation of homologous recombination-related genes is still poorly described.

The objective of this work was to evaluate the mRNA expression and promoter methylation of these genes in gastric adenocarcinoma, as well as their possible correlations with clinicopathological characteristics and with the immunohistochemical expression of their respective proteins previously published by our group.¹¹

Materials and Methods

Study population

This is a retrospective cohort study that evaluates patients already analyzed in a previous immunohistochemistry-based study of our group,¹¹ but using different methods (RT-qPCR and MS-HRM). This preceding work evaluated a larger number of patients (n = 121). The procedures and analyses of the former work were conducted between 2018 and 2020, whereas the laboratory procedures of the present study were started in 2019, interrupted in 2020 due to the COVID-19 pandemic, and completed in 2022.

Patients who underwent total or subtotal gastrectomy for gastric adenocarcinoma treatment at the Clinical Hospital of the Ribeirão Preto Medical School at the University of São Paulo (HCFMRP-USP) between April 2008 and June 2017 and who had frozen samples stored were analyzed. Only tumors infiltrating at least the muscularis propria layer were included. Patients who did not survive for at least 30 days after the surgery were excluded.

In total, 73 patients were analyzed, and their characteristics are summarized in Table 1. However, because of the availability of tumoral or non-tumoral frozen material previously stored in the biorepository, loss of samples and the presence of cases with inadequate quality of extracted material, tumor and non-tumor mucosa mRNA expression was evaluated only in 48 and 32 patients, respectively. This number of 48 cases was sufficient for evaluating the correlations between the expressions obtaining coefficients below .40, with a significance level of 5% and a power of 80%. The methylation of gene promoters was evaluated in tumor samples from 44 patients, whereas non-tumor mucosa methylation was evaluated in 49 patients. Of the 48 cases in which tumor mRNA was evaluated, 8 were treated with surgery alone; 13 received neoadjuvant treatment consisting of ECX (epirubicin, cisplatin, and capecitabine) or slight variations of this protocol; and 27 received adjuvant treatment, mainly with the MacDonald protocol or platinum-based schemes. Of these 48 patients, 31 died, with a survival time ranging from 62 to 1646 days (mean 461 ± 371). The other 17 had follow-up times ranging from 60 to 3755 days (mean 1674 ± 1242).

Table 1.

Clinicopathological characteristics of the patients.

Characteristics of the study population (N = 73)	n (%)
Age (years)	62 ± 12.8
Sex
Male	50 (68.5)
Female	23 (31.5)
Histological type
Diffuse	42 (57.5)
Intestinal	21 (28.8)
Mixed	10 (13.7)
Histological grade
1	4 (5.5)
2	13 (17.8)
3	56 (76.7)
Stage
I/II	31 (42.5)
III/IV	42 (57.5)
pT
2	14 (19.2)
3-4	59 (80.8)
Lymph node metastasis
Yes	54 (73.9)
No	19 (26.1)
Distant metastasis
Yes	13 (17.8)
No	60 (82.2)

Information on clinicopathological features—histological type, histological grade, depth of invasion, presence of regional lymph node metastasis, distant metastasis, blood or lymphatic vascular invasion, perineural invasion, stage, overall survival, and disease-free survival—was obtained from medical records and histological slides.

This study was approved by the HCFMRP-USP Research Ethics Committee (approval No. 12349/17).

This manuscript was prepared in accordance with the STROBE Guidelines.

DNA and RNA extraction

Gastric adenocarcinoma and gastric mucosa samples were collected from gastrectomy surgical specimens and stored at −80°C. Representative portions of the samples were dissected manually, using HE-stained 4 μm frozen histological sections as reference. Automated DNA extraction was performed with QiaCube (Qiagen, Germantown, MD), according to the manufacturer’s instructions. RNA extraction was performed using TRIzol (Invitrogen, Waltham, MA), following the manufacturer’s protocol.

Methylation-sensitive high-resolution melting (MS-HRM)

For methylation evaluation of the ATM, ATR, RAD51, BRCA1, and BRCA2 genes, DNA conversion using sodium bisulfite and sample cleanup of 1 μg of genomic DNA were performed using the EpiTect Bisulfite Kit (Qiagen), following the manufacturer’s instructions. Afterwards the DNA conversion, each sample was run in duplicate by the MS-HRM technique to determine the methylation pattern of the promoter regions of each gene using primers designed with the aid of MethPrimer software (http://www.urogene.org/cgi-bin/methprimer/methprimer.cgi): 5′-TTTTAATATTGAATTAAGGAAATTG-3′ and 5′-AAATACTCCTCTCTTCAAAAACTAC-3′ for ATM; 5′-TTTGTGTTTTTTAGGTTTGAGAATAGTAG-3′ and 5′-TATAAAAACCAAAACTCCTCCC-3′ for ATR; 5′-GAATTTTTGATTTTAGGTTATTTATT-3′ and 5′-TCACTATCTTAACCAAACTATTCTC-3′ for RAD51; 5′-TGGTTTTTATTATTTGTTTTTTAAAA-3′ and 5′-TCAACCCCAATATTTATTATTTTTC-3′ for BRCA1; and 5′-GGTGTGGTGGTTTATGTTTGTAAT-3′ and 5′-TCAAATAATTCTCCTACCTCAACCT-3′ for BRCA2.

MS-HRM was performed in a 7500 Fast Real-Time PCR System (Applied Biosystems) using MeltDoctor HRM Master Mix (Applied Biosystems) according to manufacturer’s instructions. The PCR reaction conditions were 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds, 58°C for 1 minutes, and 95°C for 1 minutes, followed by the standard melting curve. For the MS-HRM analysis, each sample duplicate was compared to controls, and we employed in each run two controls already treated with sodium bisulfite (Epitect PCR Control DNA Set, Qiagen), one of which was 100% methylated, and another was 100% unmethylated.

RT-qPCR

For cDNA synthesis, reverse transcription (RT) was performed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA), according to the manufacturer’s instructions. The cDNA used in the quantitative polymerase chain reaction (qPCR) was previously diluted 1:10 in water treated with diethyl pyrocarbonate (DEPC).

The following assays (ThermoFisher Scientific, Waltham, MA) were used: Hs01556193_m1 (BRCA1), Hs00609073_m1 (BRCA2), Hs00175892_m1 (ATM), Hs00992123_m1 (ATR), Hs00153418_m1 (RAD51), Hs99999903_m1 (ACTB), and Hs99999907_m1 (B2M). qPCR amplification reactions were performed in duplicates using Taqman Master Mix (Applied Biosystems, Foster City, CA). Each reaction had a final volume of 10 μl, using 5 μl of the specific Taqman Master Mix reagent, 0.5 μl of each specific probe, and 4.5 μl of diluted cDNA. The amplification conditions were 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minutes (simultaneous annealing and extension). The real-time PCR detection device 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA) was used together with 7500 Sequence Detection System software (Applied Biosystems, Foster City, CA) to obtain Ct values.

The expression levels (fold changes) were given by the formula 2^{(−∆∆Ct)}, where ∆Ct was the difference between the Ct of the studied gene and the mean Ct of the endogenous ACTB and B2M for the same sample, and ∆∆Ct was the difference between the ∆Ct of the sample and the mean ∆Ct of non-neoplastic samples. The absence of mRNA amplification of the gene under study, combined with preserved amplification of the endogenous controls, was considered the absence of expression. A fold change of 0 was assigned to cases without gene expression, only for statistical correlation with other variables.

Immunohistochemistry

Immunohistochemistry data that are part of an earlier study published by our group were used for correlations with RT-qPCR results.¹¹ The following antibodies were used: ATM (St John’s Laboratory, London, UK, STJ97797, 1:400), ATR (Abcam, Cambridge, UK, ab178407, 1:100), BRCA1 (St John’s Laboratory, STJ113833, 1:300), RAD51 (St John’s Laboratory, STJ95330, 1:100), and BRCA2 (St John’s Laboratory, STJ91885, 1:100). As previously described, immunohistochemistry results were expressed as histochemical scores (H-scores) and were additionally classified as positive or negative. The H-score was calculated by multiplying the percentage of stained cells by 1, 2, or 3 for weak, moderate, or strong staining, respectively. Cases with 10% or less stained neoplastic cells were considered negative; those with more than 10% were considered positive.

Statistical analyses

The Shapiro-Wilk test was used to study the normality of the data. The non-parametric Mann-Whitney test was used for comparing two groups, and the non-parametric Kruskal-Wallis test was used for comparing three groups, because the assumption of normality of the data was rejected. For comparing tumor and non-tumor samples, Wilcoxon’s non-parametric test was used, because the assumption of normality of the data was rejected. Spearman’s correlation coefficient was used to study correlations. The ROC curve was used in an attempt to define the expression cut-off point to predict the occurrence of death. The chi-square test verified homogeneity between the proportions. Survival was studied using Kaplan-Meier survival curves with log-rank tests and the univariate Cox model. Kaplan-Meier curves were tested with all possible cut-off points to define low and high expression. The significance level was 5%. The software used for the calculations was SPSS 17.0 for Windows. To avoid bias, additional clinicopathological correlation analyzes were carried out excluding cases with neoadjuvant treatment and, in the case of death and survival assessment, with other groupings according to the type of treatment.

Results

Methylation in tumors and non-tumoral mucosa

Considering the tumor samples, total methylation of RAD51 was detected in 7.7% of the cases (3/39), partial methylation was detected in 87.2% (34/39), and absence of methylation, in 5.1% (2/39); total BRCA1 methylation was detected in 2.4% of the cases (1/41), partial methylation was verified in 95.1% (39/41), and absence of methylation, in 2.4% (1/41); total BRCA2 methylation was seen in all cases (41/41); whereas the absence of ATR (42/42) and ATM (40/40) methylation was seen in all cases.

In non-tumor samples, total methylation of RAD51 was detected in 6.1% of the cases (3/49), and partial methylation was detected in 93.9% (46/49); total BRCA1 methylation was detected in 2.1% of the cases (1/48), partial methylation was verified in 95.8% of the cases (46/48), and absence of methylation was observed in 2.1% (1/48); total BRCA2 methylation was verified in all cases (49/49); and absence of ATR (49/49) and ATM (49/49) methylation was seen in all cases.

There was agreement in the methylation status of paired tumor and non-tumor samples in 78.4% (29/37) of the cases for RAD51, and in 94.6% (35/37) of the cases for BRCA1 (Table 2). However, it was not possible to apply statistical tests to assess the significance of this correlation because of the great predominance of partial methylation.

Table 2.

Cross-analysis of the methylation of RAD51 and BRCA1 genes in tumor and non-tumor mucosa, considering the 37 cases with paired tumoral and non-tumoral samples.

	Tumoral	Non-tumoral (%)
		Absent	Partial	Total
RAD51	absent	–	2 (5.4)	0 (0.0)
	partial	–	29 (78.4)	3 (8.1)
	total	–	3 (8.1)	0 (0.0)
BRCA1	absent	0 (0.0)	1 (2.7)	0 (0.0)
	partial	1 (2.7)	34 (91.9)	0 (0.0)
	total	0 (0.0)	0 (0.0)	1 (2.7)

Similarly, because of the qualitative nature of the methylation assessment and the reduced number of divergent results, it was not possible to statistically evaluate the correlation between methylation and other variables.

mRNA expression in tumors and non-tumoral mucosa

In tumor samples, the absence of RAD51, ATR, BRCA1, BRCA2, and ATM mRNA expression was detected in 6.25%, 6.25%, 8.33%, 16.67%, and 10.42% of the cases, respectively. More detailed results are shown in Table 3.

Table 3.

RAD51, ATR, BRCA1, BRCA2, and ATM mRNA expression in gastric adenocarcinoma and non-tumor gastric mucosa samples.

	mRNA expression		mRNA fold change in cases with expression
RAD51
Tumor	Absent	6.25% (3/48)
	Present	93.75% (45/48)	Mean 2.0760
			min: 0.1429
			max: 8.6607
			SD: 2.1142
Non-tumor mucosa	Absent	31.25% (10/32)
	Present	68.75% (22/32)	Mean: 1.6572
			min: 0.0283
			max: 7.6339
			SD: 1.7572
ATR
Tumor	Absent	6.25% (3/48)
	Present	93.75% (45/48)	Mean: 1.4332
			min: 0.0871
			max: 17.0522
			SD: 2.6250
Non-tumor mucosa	Absent	32.26% (10/31)
	Present	67.74% (21/31)	Mean: 5.2802
			min: 0.1427
			max: 83.8923
			SD: 18.0676
BRCA1
Tumor	Absent	8.33% (4/48)
	Present	91.67% (44/48)	Mean: 0.6419
			min: 0.0266
			max: 2.8812
			SD: 0.6753
Non-tumor mucosa	Absent	18.75% (6/32)
	Present	81.25% (26/32)	Mean: 2.1834
			min: 0.0328
			max: 9.2877
			SD: 2.3584
BRCA2
Tumor	Absent	16.67% (8/48)
	Present	83,33% (40/48)	Mean: 0.4695
			min: 0.0108
			max: 4.3376
			SD: 0.8572
Non-tumor mucosa	Absent	43.75% (14/32)
	Present	56.25% (18/32)	Mean: 23.9100
			min: 0.0016
			max: 270.0019
			SD: 65.9849
ATM
Tumor	Absent	10.42% (5/48)
	Present	89.58% (43/48)	Mean: 10.3642
			min: 0.4420
			max: 76.1618
			SD: 14.2744
Non-tumor mucosa	Absent	34.37% (11/32)
	Present	65.62% (21/32)	Mean: 7.0627
			min: 0.3872
			max: 33.9450
			SD: 7.7460

Proportion of cases with no expression and with present expression, mean, minimum (min.), and maximum (max.) fold change values and standard deviations (SD) are shown, calculated considering the cases with present expression.

Considering the paired samples, non-tumor mucosa showed significantly higher median level of BRCA1 mRNA than tumors (1.272 vs 0.345, P = .015). No significant difference was observed for the other genes (Figure 1).

Figure 1.

Comparison of RAD51, ATR, BRCA1, BRCA2, and ATM mRNA expression in tumor and non-tumor mucosa, considering the paired samples of the same individuals. The results are presented in box plots, using a logarithmic scale (+1 was added as a constant to allow the logarithmic representation of null values).

The tumor and non-tumor mRNA expression of the studied genes was not statistically different in cases with previous neoadjuvant treatment compared with the other cases (Table 4).

Table 4.

Comparison between RAD51, ATR, BRCA1, BRCA2, and ATM mRNA expression in patients with previous neoadjuvant chemotherapy and in those without neoadjuvant chemotherapy, considering tumor samples and non-tumor mucosa samples.

	Neoadjuvant therapy	n	Median mRNA fold change	P
RAD51
Tumor	Yes	13	1.250	.307
Tumor	No	35	1.664	.307
Non-tumor	Yes	14	0.585	.749
Non-tumor	No	18	0.865	.749
ATR
Tumor	Yes	13	0.330	.194
Tumor	No	35	0.693	.194
Non-tumor	Yes	14	0.349	.749
Non-tumor	No	18	0.482	.749
BRCA1
Tumor	Yes	13	0.217	.298
Tumor	No	35	0.520	.298
Non-tumor	Yes	14	0.598	.497
Non-tumor	No	18	1.231	.497
BRCA2
Tumor	Yes	13	0.121	.928
Tumor	No	35	0.081	.928
Non-tumor	Yes	14	0.001	.849
Non-tumor	No	18	0.069	.849
ATM
Tumor	Yes	13	2.998	.057
Tumor	No	35	4.811	.057
Non-tumor	Yes	14	0.927	.255
Non-tumor	No	18	2.011	.255

There were positive and significant weak to moderate correlations between tumor mRNA expression levels of RAD51 and ATR (r = .319, P = .027), RAD51 and BRCA1 (r = .587, P < .001), RAD51 and BRCA2 (r = .502, P < .001), ATR and BRCA1 (r = .383, P = .007), ATR and ATM (r = .449, P = .001), and BRCA1 and BRCA2 (r = .474, P = .001).

Immunohistochemistry and mRNA correlation

Tumors that were positive for nuclear ATR in the immunohistochemical evaluation (more than 10% of neoplastic cells with nuclear staining) had a lower median mRNA fold change than those that were negative (0.487 vs 0.907, P = .032). The other immunohistochemical markers showed no significant correlation (Table 5).

Table 5.

Association between immunohistochemistry and mRNA levels in tumor samples.

Immunohistochemistry	n	Median mRNA fold-change	P	r (considering H-score)	P
Nuclear RAD51
Positive	38	1.324	.317	−0.136	.361
Negative	9	1.286			.361
Cytoplasmic RAD51
Positive	35	1.359	.726	0.168	.260
Negative	12	1.226			.260
Nuclear ATR
Positive	18	0.487	.032	−0.279	.055
Negative	30	0.907			.055
Cytoplasmic ATR
Positive	11	0.651	.826	−0.114	.442
Negative	37	0.602			.442
Nuclear BRCA1
Positive	42	0.361	.267	−0.069	.643
Negative	6	0.583			.643
Cytoplasmic BRCA1
Positive	43	0.390	.897	−0.027	.855
Negative	5	0.413			.855
Nuclear BRCA2
Positive	21	0.124	.624	0.124	.408
Negative	27	0.056			.408
Cytoplasmic BRCA2
Positive	13	0.173	.384	0.116	.434
Negative	35	0.081			.434
Nuclear ATM
Positive	43	4.638	NA	−0.014	.927
Negative	4	6.541			.927

Two different analyses are shown: one considers the immunohistochemistry results classified into two categories—positive or negative—, and the other considers the Spearman correlation coefficients between H-scores and mRNA fold changes. Bold indicates statistical significance.

mRNA and clinicopathological parameters

Cases with perineural invasion had higher median tumor amounts of RAD51 mRNA (1.822 vs 0.725, P = .010). Size, histological type, histological grade, vascular invasion, depth of invasion, lymph node metastasis, distant metastasis, and stage did not correlate with gene expression in the total population (Tables 6 and 7). Given the possibility of an influence of previous neoadjuvant chemotherapy on mRNA expression, all analyses were also performed excluding patients who received neoadjuvant treatment. This showed that deeper infiltrating tumors (pT3–4, n = 30) had lower median levels of BRCA1 mRNA than superficial tumors (pT2, n = 5)—0.414 versus 0.954, P = .035. The association between high median RAD51 mRNA and perineural invasion was also maintained in this group (2.028 in the presence of perineural invasion vs 0.456 in the absence of perineural invasion, P = .024).

Table 6.

Correlation between RAD51, ATR, BRCA1, BRCA2, and ATM mRNA expression and histological type, histological grade, perineural invasion, and vascular invasion in tumor samples.

Histological type	n	mRNA fold-change median	P	Histological grade	n	mRNA fold-change median	P	Perineural invasion	n	mRNA fold-change median	P	Vascular invasion	n	mRNA fold-change median	P
RAD51
Diffuse	31	0.977	.205	1-2	7	2.453	.077	No	24	0.725	.010	No	11	1.908	.254
Intestinal	9	1.908		3	41	1.165		Yes	24	1.822		Yes	37	1.165
Mixed	8	1.570
ATR
Diffuse	31	0.581	0.926	1-2	7	1.114	.140	No	24	0.663	.773	No	11	1.062	.173
Intestinal	9	1.062		3	41	0.581		Yes	24	0.627		Yes	37	0.581
Mixed	8	0.781
BRCA1
Diffuse	31	0.259	0.590	1-2	7	0.517	.261	No	24	0.344	.606	No	11	0.519	.291
Intestinal	9	0.517		3	41	0.337		Yes	24	0.524		Yes	37	0.278
Mixed	8	0.542
BRCA2
Diffuse	31	0.063	0.196	1-2	7	0.526	.121	No	24	0.074	.542	No	11	0.327	.113
Intestinal	9	0.327		3	41	0.077		Yes	24	0.131		Yes	37	0.066
Mixed	8	0.167
ATM
Diffuse	31	4.811	0.389	1-2	7	4.228	.942	No	24	4.825	.433	No	11	4.483	.797
Intestinal	9	2.711		3	41	4.811		Yes	24	4.450		Yes	37	4.638
Mixed	8	8.677

Bold indicates statistical significance.

Table 7.

Correlation between RAD51, ATR, BRCA1, BRCA2, and ATM mRNA expression and depth of invasion (pT), regional lymph node metastasis, distant metastasis, and stage in tumor samples.

	pT	N	mRNA fold-change median	P	Lymph node metastasis	N	mRNA fold-change median	P	Distant metastasis	n	mRNA fold-change median	P	Stage	n	mRNA fold-change median	P
RAD51	2	39	1.019	.607	No	13	1.822	.379	No	39	1.288	.802	III-IV	29	1.286	.712
RAD51	3-4	9	1.359	.607	Yes	34	1.225	.379	Yes	9	1.286	.802	I-II	19	1.811	.712
ATR	2	39	0.994	.663	No	13	1.062	.207	No	39	0.651	.782	III-IV	29	0.651	.891
ATR	3-4	9	0.602	.663	Yes	34	0.592	.207	Yes	9	0.536	.782	I-II	19	0.581	.891
BRCA1	2	39	0.572	.122	No	13	0.517	.468	No	39	0.491	.663	III-IV	29	0.259	.158
BRCA1	3-4	9	0.298	.122	Yes	34	0.307	.468	Yes	9	0.203	.663	I-II	19	0.519	.158
BRCA2	2	39	0.081	.361	No	13	0.259	.105	No	39	0.124	.397	III-IV	29	0.063	.311
BRCA2	3-4	9	0.124	.361	Yes	34	0.065	.105	Yes	9	0.046	.397	I-II	19	0.144	.311
ATM	2	39	2.997	.362	No	13	4.483	.757	No	39	4.638	.335	III-IV	29	4.811	.744
ATM	3-4	9	4.811	.362	Yes	34	4.724	.757	Yes	9	4.417	.335	I-II	19	4.483	.744

The occurrence of death correlated with higher tumoral RAD51 mRNA (1.664 vs 0.929, P = .036), even if cases with neoadjuvant treatment were excluded (Table 8). However, we were not able to define an expression cut-off point that could be used as a statistically significant predictor of death. Moreover, using the Cox model, we observed no association between overall survival and tumor mRNA expression levels of RAD51 (P = .510), ATR (P = .893), BRCA1 (P = .310), BRCA2 (P = .488), and ATM (P = .153). There was also no association between disease-free survival and RAD51 (P = .271), ATR (P = .474), BRCA1 (P = .566), BRCA2 (P = .288), and ATM (P = .230) expression. Separate analyses according to the type of treatment (surgery alone, adjuvant, neoadjuvant, adjuvant + neoadjuvant, and adjuvant + surgery alone) similarly did not show statistically significant correlations. Figure 2 illustrates the absence of a significant correlation between tumor mRNA expression and overall survival.

Table 8.

Comparison between the mRNA levels in the tumors of patients who died and those who did not, considering the following groups: total population, excluding neoadjuvant therapy, with only surgical treatment, and with adjuvant or neoadjuvant treatment.

Total population	Death	n	mRNA fold-change median	P	Only surgery	Death	n	mRNA fold-change median	P
RAD51	No	17	0.929	.036	RAD51	No	3	0.457	n.a.
RAD51	Yes	31	1.664	.036	RAD51	Yes	5	3.406	n.a.
ATR	No	17	0.651	.575	ATR	No	3	0.994	n.a.
ATR	Yes	31	0.602	.575	ATR	Yes	5	0.305	n.a.
BRCA1	No	17	0.413	.872	BRCA1	No	3	0.521	n.a.
BRCA1	Yes	31	0.337	.872	BRCA1	Yes	5	0.645	n.a.
BRCA2	No	17	0.081	.522	BRCA2	No	3	0.017	n.a.
BRCA2	Yes	31	0.124	.522	BRCA2	Yes	5	0.144	n.a.
ATM	No	17	4.811	.548	ATM	No	3	9.397	n.a.
ATM	Yes	31	4.483	.548	ATM	Yes	5	4.638	n.a.
Without neoadjuvant					Adjuvant + neoadjuvant
RAD51	No	10	0.405	.018	RAD51	No	14	0.974	.096
RAD51	Yes	25	1.855	Yes	RAD51	26	1.394	0.974	.096
ATR	No	10	0.823	.596	ATR	No	14	0.539	.992
ATR	Yes	25	0.693	Yes	ATR	26	0.647	0.539	.992
BRCA1	No	10	0.520	.674	BRCA1	No	14	0.401	.535
BRCA1	Yes	25	0.517	.674	BRCA1	Yes	26	0.238	.535
BRCA2	No	10	0.048	.083	BRCA2	No	14	0.130	.968
BRCA2	Yes	25	0.142	.083	BRCA2	Yes	26	0.100	.968
ATM	No	10	4.179	.289	ATM	No	14	3.272	.218
ATM	Yes	25	6.782	.289	ATM	Yes	26	4.450	.218

Bold indicates statistical significance.

Figure 2.

Kaplan-Meier curves comparing the overall survival of the 24 patients with tumor mRNA expression above the median with that of the 24 below the median, for RAD51, ATR, BRCA1, BRCA2, and ATM.

Discussion

Although the incidence of gastric cancer has decreased in recent decades, it is still the fifth cancer in incidence and the fourth in mortality worldwide.¹ Previous studies have found an association between altered immunohistochemical expression of homologous recombination proteins and clinicopathological features of gastric adenocarcinoma.^8-18

Specifically concerning the BRCA1 protein, such studies had discrepant results regarding its cellular location—nuclear,^8,9 cytoplasmic,¹⁰ or both.^11-13 The correlation with clinicopathological parameters is also controversial. It was absent in some studies,^11,13 and, in others, low expression was associated with advanced tumors and worse prognosis.^8-10 For Wang et al,¹² not only specifically low cytoplasmic expression but also high nuclear expression is associated with unfavorable characteristics.

Here, we showed that tumors with deeper infiltration of the gastric wall (subserosa or beyond—T3-T4) have lower levels of BRCA1 mRNA than those limited to the muscularis propria (T2) in the group of patients who had not received previous neoadjuvant therapy. We also demonstrated reduced BRCA1 mRNA in adenocarcinoma compared with non-tumor mucosa. Regarding survival, although there seems to be a tendency toward greater survival in cases with greater expression, this was not statistically significant. These results, together with some of the previously published data, support the hypothesis that a decrease in BRCA1 expression favors the development or progression of gastric cancer.

In contrast, Kim et al¹³ did not observe any clinicopathological association of BRCA1 mRNA expression in gastric cancer.

According to Zhang et al,⁹ the immunohistochemical expression of ATR has no clinicopathological associations. The previous study from our group, in contrast, found a correlation between negative nuclear immunohistochemical expression of ATR and higher histological grade and tumor size in gastric adenocarcinoma.¹¹ Here, however, there was no correlation between ATR mRNA and clinicopathological characteristics. Furthermore, we observed an inverse relationship between immunohistochemical results and mRNA levels, as immunohistochemically positive cases for nuclear ATR had a significantly lower median mRNA fold change than negative cases. This suggests that there may be an effect of post-transcriptional or post-translational regulation on ATR protein expression. In renal carcinoma cell cultures, for example, miR-185 inhibits ATR protein synthesis.²⁰ In pancreatic cancer cells, ATR undergoes degradation by the E3 ubiquitin ligase FBXO32.²¹

Studies on the prognostic importance of RAD51 in various types of cancer have had conflicting results. Although this protein acts on DNA repair, having therefore certain tumor suppressor character, its overexpression may lead to genomic instability and carcinogenesis.²² It is also known that RAD51 increases the expression of the EBPβ transcription factor, which favors the epithelial-mesenchymal transition and the synthesis of metalloproteinases.²³ An association between poor prognosis and low immunohistochemical RAD51 expression was noted in glioblastoma,²⁴ breast carcinoma,^25,26 and non-small cell lung cancer.²⁷ Conversely, correlation between RAD51 and unfavorable characteristics such as high histological grade, lymph node metastases, or lower survival was described in colon cancer,^28,29 breast carcinoma,^23,30 esophageal squamous cell carcinoma,³¹ head and neck tumors,³² non-small cell lung cancer,³³ and prostatic adenocarcinoma.³⁴

Specifically in gastric cancer, our previous published data have shown that immunohistochemically negative cases for nuclear RAD51 had a worse prognosis, a tendency of lymphatic dissemination, and a larger mean tumor size.¹¹ These cases, however, responded better to chemoradiotherapy.^11,15

However, in contrast with those previous immunohistochemical findings, in this study, RAD51 mRNA levels were unexpectedly higher in cases with perineural invasion than in other cases, as well as in the tumors of patients who had died. This may indicate a possible complex role of RAD51 in the pathogenesis of gastric cancer.

Despite this correlation between RAD51 and the occurrence of death, and although the survival of cases with high expression appears to be lower in the Kaplan-Meier curve, we were unable to demonstrate a statistically significant association with overall or disease-free survival. This may be due to the small number of patients evaluated in this study. It is necessary to consider that a hypothetical association between RAD51 mRNA and lower survival could be due to resistance to chemoradiotherapy. Our number of untreated patients was too small to allow for reliable statistical analysis.

We also need to consider the possibility of an actual absence of correlation between the nuclear presence of the protein and the amount of mRNA. In fact, we noticed that RAD51 mRNA even had a slight negative correlation with the nuclear H-score and a positive correlation with the cytoplasmic H-score, both of which were not statistically significant. Nuclear localization of RAD51 depends on a complex cytoplasmic-nuclear transport mechanism mediated by BRCA2 and RAD51C.³⁵ Similar to ATR expression, RAD51 protein synthesis may be suppressed by post-transcriptional regulation, as demonstrated by Huang et al³⁶ in neoplastic cell cultures in which miR-103 and miR-107 inhibited RAD51 expression and the occurrence of homologous recombination. In addition, the microRNAs miR-182, miR-221, miR-34a, and miR-766 inhibit the synthesis of the RAD51 protein.³⁷ Additionally, RAD51 is ubiquitinated and undergoes degradation by FBH1, FBXO5, and RFWD3.³⁸

In contrast to our finding, Redati et al³⁷ noted that high levels of RAD51 mRNA are associated with greater survival in gastric adenocarcinoma and esophageal squamous cell carcinoma, but with a worse prognosis in hepatocellular carcinoma and esophageal, breast, lung, and gastroesophageal junction adenocarcinomas.

The lack of a significant association between immunohistochemistry and mRNA expression, as observed for RAD51, ATM, BRCA1, and BRCA2, may also be due to the semiquantitative and subjective nature of the immunohistochemistry analysis, especially when staining intensity is used to calculate the H-score, and to its qualitative and dualistic character when only negative or positive categories are used. Kim et al¹³ similarly did not observe a correlation between immunohistochemical results and BRCA1 or BRCA2 mRNA expression in gastric cancer.

Previous immunohistochemical studies suggest that high nuclear expression^11,14 or low cytoplasmic expression^12,13 of BRCA2 and low expression of ATM^11,16-18 are characteristics of more advanced gastric tumors and are associated with shorter survival. In the present study, however, we did not observe a correlation between ATM or BRCA2 mRNA levels and clinicopathological parameters or with immunohistochemistry. Likewise, Kim et al¹³ did not notice any association between BRCA2 mRNA levels and clinicopathological characteristics.

BRCA1 promoter methylation was a frequent finding in the work by Bernal et al,³⁹ who evaluated diffuse-type adenocarcinomas. We detected a high proportion of cases with complete or partial methylation of BRCA1, BRCA2, and RAD51, which suggests that this mechanism participates in regulating the expression of these genes in gastric cancer, but not of ATM and ATR, where promoter methylation was not detected. In contrast, ATM promoter hypermethylation has been observed in breast cancer, gliomas, gastric lymphoma, and colorectal tumors.⁴⁰

Additional findings of this work were the positive weak to moderate correlations between tumor mRNA expression levels of RAD51 with ATR, BRCA1, and BRCA2; of ATR with BRCA1 and ATM; and of BRCA1 with BRCA2. This suggests that there may be a common control of the mRNA synthesis of such genes, possibly occurring in response to DNA damage.

In summary, our findings suggest that alterations in BRCA1 and RAD51 mRNA expression play a role in the progression of gastric adenocarcinoma, although, in the case of RAD51, there seems to be a discrepancy between the clinical significance of mRNA expression and that of the previously described nuclear immunohistochemical positivity of the protein. Additional studies are needed to define whether there would be prognostic significance in a larger population and to elucidate the mechanisms that influence the relationship between protein and mRNA expression.

The main limitation of this study is the small number of patients analyzed, which did not allow for a definitive assessment of the possible prognostic implications of RAD51 and BRCA1, especially if we consider that they were subjected to different therapeutic modalities. Furthermore, the number of cases in which it was possible to analyze paired tumor and non-tumor expression was limited. Other limitations were the qualitative nature of the method used to evaluate gene methylation, and the possible influence of tumor heterogeneity in the attempt to correlate immunohistochemistry and mRNA levels, as the samples were from different areas of the tumor.

Conclusions

RAD51 mRNA high expression in gastric adenocarcinoma was associated with perineural invasion and death. BRCA1 mRNA was reduced in tumors compared with non-neoplastic mucosa and, excluding neoadjuvant therapy cases, in deeper infiltrating tumors. Methylation of the RAD51, BRCA1, and BRCA2 promoters was highly frequent, whereas that of ATM and ATR was not detected. ATR nuclear immunohistochemical positivity and mRNA levels were inversely correlated, whereas there was no significant mRNA and immunohistochemistry correlation for the other markers.

Notation

List of Abbreviations

ACTB beta-actin gene

ATM ataxia-telangiectasia mutated

ATR ataxia-telangiectasia and Rad3-related

BRCA1 breast cancer 1

BRCA2 breast cancer 2

B2M beta-2 microglobulin gene

cDNA complementary DNA

Ct cycle threshold

DEPC diethyl pyrocarbonate

EBPβ CCAAT-enhancer-binding protein beta

ECX epirubicin, cisplatin, and capecitabine

FBH1 F-box DNA helicase 1

FBXO5 F-box only protein 5

FBXO32 F-box only protein 32

HCFMRP-USP Clinical Hospital of the Ribeirão Preto Medical School at the University of São Paulo

HE hematoxylin and eosin

H-score histochemical score

miR microRNA

mRNA messenger RNA

MS-HRM methylation-sensitive high-resolution melting

RAD51 radiation-sensitive protein 51

RFWD3 ring finger and WD repeat domain 3

ROC receiver operating characteristic

RT-qPCR reverse transcription-quantitative polymerase chain reaction

Footnotes

Acknowledgements

We acknowledge Creusa Maria Roveri Dal Bó for statistical support, and Fábio Antônio Venâncio for graph construction.

Declarations

ORCID iD

Joel Del Bel Pádua

References

Morgan

Arnold

Camargo

, et al. The current and future incidence and mortality of gastric cancer in 185 countries, 2020–40: a population-based modelling study. EClinicalMedicine 2022;47:101404.

Fukayama

Rugge

Washington

. Tumours of the stomach. In: World Health Organization (ed.) WHO Classification of Tumours: Digestive System Tumours, Vol. 1. 5th ed. IARC; 2019;59-109.

Handa

Naito

Yoshikawa

Redox biology and gastric carcinogenesis: the role of Helicobacter pylori. Redox Rep. 2011;16:1-7.

Hanada

Uchida

Tsukamoto

, et al. Helicobacter pylori infection introduces DNA double-strand breaks in host cells. Infect Immun. 2014;82:4182-4189.

Foray

Marot

Gabriel

, et al. A subset of ATM- and ATR-dependent phosphorylation events requires the BRCA1 protein. EMBO J. 2003;22:2860-2871.

Forget

Kowalczykowski

. Single-molecule imaging brings Rad51 nucleoprotein filaments into focus. Trends Cell Biol. 2010;20:269-276.

Heeke

Pishvaian

Lynce

, et al. Prevalence of homologous recombination-related gene mutations across multiple cancer types. JCO Precis Oncol. 2018;2018:PO.17.00286.

Kim

Cho

Kim

, et al. Differing effects of adjuvant chemotherapy according to BRCA1 nuclear expression in gastric cancer. Cancer Chemother Pharmacol. 2013;71:1435-1443.

Zhang

Liu

Yin

, et al. Loss of BRCA1 expression leads to worse survival in patients with gastric carcinoma. World J Gastroenterol. 2013;19:1968-1974.

10.

Chen

Wang

, et al. Prognostic significance of BRCA1 expression in gastric cancer. Med Oncol. 2013;30:423.

11.

Padua

JDB

Mariano

CFA

Fabro

, et al. Prognostic value of the immunohistochemical expression of RAD51 and BRCA2 in gastric adenocarcinoma. J Histochem Cytochem. 2022;70:199-210.

12.

Wang

Zhao

Huang

, et al. BRCA1 and BRCA2 expression patterns and prognostic significance in digestive system cancers. Hum Pathol. 2018;71:135-144.

13.

Kim

Hwang

Min

, et al. Clinical significance of BRCA1 and BRCA2 mRNA and protein expression in patients with sporadic gastric cancer. Oncol. Lett. 2019;17:4383-4392.

14.

Zhang

Yang

, et al. Prognostic value of the expression of DNA repair-related biomarkers mediated by alcohol in gastric cancer patients. Am J Pathol. 2018;188:367-377.

15.

Tan

Yuan

Zhu

, et al. Interferon regulatory factor-1 suppresses DNA damage response and reverses chemotherapy resistance by downregulating the expression of RAD51 in gastric cancer. Am J Cancer Res. 2020;10:1255-1270.

16.

Abdel-Fatah

Arora

Gorguc

, et al. Are DNA repair factors promising biomarkers for personalized therapy in gastric cancer? Antioxid Redox Signal. 2013;18:2392-2398.

17.

Lee

Han

Kim

, et al. DNA damage response-related proteins in gastric cancer: ATM, Chk2 and p53 expression and their prognostic value. Pathobiology. 2014;81:25-35.

18.

Kim

, et al. Ataxia-telangiectasia-mutated protein expression with microsatellite instability in gastric cancer as prognostic marker. Int J Cancer. 2014;134:72-80.

19.

Zeng

Rong

, et al. DNA methylation: an important biomarker and therapeutic target for gastric cancer. Front. Genet. 2022;13:823905.

20.

Wang

, et al. Repression of ATR pathway by miR-185 enhances radiation-induced apoptosis and proliferation inhibition. Cell Death Dis. 2013;4:e699.

21.

Yang

Fan

Zhu

, et al. 3F-Box protein 32 degrades ataxia telangiectasia and Rad3-related and regulates DNA damage response induced by gemcitabine in pancreatic cancer. Oncol Lett. 2018;15:8878-8884.

22.

Klein

HL.

The consequences of Rad51 overexpression for normal and tumor cells. DNA Repair (Amst). 2008;7:686-693.

23.

Wiegmans

Al-Ejeh

Chee

, et al. Rad51 supports triple negative breast cancer metastasis. Oncotarget. 2014;5:3261-3272.

24.

Welsh

Ellsworth

Kumar

, et al. Rad51 protein expression and survival in patients with glioblastoma multiforme. Int J Radiat Oncol Biol Phys. 2009;74:1251-1255.

25.

Alshareeda

Negm

Aleskandarany

, et al. Clinical and biological significance of RAD51 expression in breast cancer: a key DNA damage response protein. Breast Cancer Res Treat. 2016;159:41-53.

26.

Soderlund

Skoog

Fornander

, et al. The BRCA1/BRCA2/Rad51 complex is a prognostic and predictive factor in early breast cancer. Radiother Oncol. 2007;84:242-251.

27.

Gachechiladze

Skarda

Kolek

, et al. Prognostic and predictive value of loss of nuclear RAD51 immunoreactivity in resected non-small cell lung cancer patients. Lung Cancer. 2017;105:31-38.

28.

Tennstedt

Fresow

Simon

, et al. RAD51 overexpression is a negative prognostic marker for colorectal adenocarcinoma. Int J Cancer. 2013;132:2118-2126.

29.

Wang

Xiao

, et al. Overexpression of Rad51 predicts poor prognosis in colorectal cancer: our experience with 54 patients. PLoS One. 2017;12:e0167868.

30.

Maacke

Opitz

Jost

, et al. Over-expression of wild-type Rad51 correlates with histological grading of invasive ductal breast cancer. Int J Cancer. 2000;88:907-913.

31.

Nakanoko

Saeki

Morita

, et al. Rad51 expression is a useful predictive factor for the efficacy of neoadjuvant chemoradiotherapy in squamous cell carcinoma of the esophagus. Ann Surg Oncol. 2014;21:597-604.

32.

Connell

Jayathilaka

Haraf

, et al. Pilot study examining tumor expression of RAD51 and clinical outcomes in human head cancers. Int J Oncol. 2006;28:1113-1119.

33.

Qiao

Yang

, et al. High-level expression of Rad51 is an independent prognostic marker of survival in non-small-cell lung cancer patients. Br J Cancer. 2005;93:137-143.

34.

Mitra

Jameson

Barbachano

, et al. Overexpression of RAD51 occurs in aggressive prostatic cancer. Histopathology. 2009;55:696-704.

35.

Gildemeister

Sage

Knight

KL.

Cellular redistribution of Rad51 in response to DNA damage: novel role for Rad51C. J Biol Chem. 2009;284:31945-31952.

36.

Huang

Wang

Dhillon

, et al. Systematic screen identifies miRNAs that target RAD51 and RAD51D to enhance chemosensitivity. Mol Cancer Res. 2013;11:1564-1573.

37.

Redati

Yang

Lei

Liu

Wang

Expression and prognostic value of RAD51 in adenocarcinoma at the gastroesophageal junction. Iran J Public Health. 2022;51:2231.

38.

Orhan

Velazquez

Tabet

, et al. Regulation of RAD51 at the transcriptional and functional levels: what prospects for cancer therapy? Cancers (Basel). 2021;13:2930.

39.

Bernal

Vargas

Ossandon

, et al. DNA methylation profile in diffuse type gastric cancer: evidence for hypermethylation of the BRCA1 promoter region in early-onset gastric carcinogenesis. Biol Res. 2008;41:303-315.

40.

Delmonico

Silva Magalhaes Costa

Gomes

, et al. Methylation profiling in promoter sequences of ATM and CDKN2A (p14(ARF)/p16(INK4a)) genes in blood and cfDNA from women with impalpable breast lesions. Oncol Lett. 2020;19:3003-3010.