RRM1 expression is associated with the outcome of gemcitabine-based treatment of non-small cell lung cancer patients–a short report

Abstract

Purpose

RRM1 is the large subunit of ribonucleotide reductase (RNR), which catalyzes the rate-limiting step in the production of deoxyribonucleotides (dNTPs) and is essential for DNA synthesis and repair. Through a meta-analysis of observational studies, we evaluated whether RRM1 expression levels are associated with the clinical outcome of gemcitabine-containing treatment regimens in patients with advanced non-small cell lung cancer (NSCLC).

Methods

A literature search was conducted using the PubMed, Embase, Web of Science, Wanfang and Chinese National Knowledge Infrastructure databases from inception to September 2014. A meta-analysis was conducted to pool eligible studies, and pooled analyses were performed using fixed effects models.

Results

A total of 12 studies encompassing 593 NSCLC patients met our search criteria and were, therefore, included. Pooled analyses revealed that patients with low/negative RRM1 expression levels exhibited significantly higher response rates (OR: 0.35, 95 % CI 0.24–0.51) and better survival rates (OR: 0.41, 95 % CI 0.23–0.75) than those with high/positive RRM1 expression levels. Subgroup analyses did not reveal any significant heterogeneity in outcome regarding the RRM1 assessment methods used or the ethnicities of patient populations studied.

Conclusions

The meta-analysis reported here indicates that RRM1 expression is associated with the response rate and overall survival rate of advanced NSCLC patients treated with gemcitabine-based chemotherapy. Additional phase III randomized trials are required to confirm our current findings.

Introduction

Lung cancer is the leading cause of cancer-related mortality worldwide, with many patients presenting with advanced disease at the time of diagnosis [1–4]. Non-small cell lung cancer (NSCLC) is the most common form of lung cancer. Platinum-based doublet chemotherapy is presently the standard first-line treatment regimen for advanced NSCLC patients. Gemcitabine with platinum derivatives or new-generation anticancer agents are being widely used in advanced NSCLC. Drug resistance is, however, an important cause of treatment failure in lung cancer and is associated with a poor 5-year survival rate of 14 % [5–7]. In order to improve the efficacy of treatment regimens in advanced NSCLC, it is necessary to stratify patients into likely responders and non-responders to specific chemotherapeutic strategies.

Ribonucleotide reductase (RNR) catalyzes the rate-limiting step in the production of deoxyribonucleotides (dNTPs) from ribonucleotides [8]. Mammalian RNR is composed of two identical large RRM1 subunits and two small subunits of either RRM2 or p53R2. Accumulating evidence indicates that increased RNR expression levels and/or activity levels can promote tumorigenesis and may contribute to resistance to chemotherapy and/or radiotherapy. Indeed, numerous studies have demonstrated that increased expression of the RRM1 subunit is associated with a poor response of cancer patients to chemotherapeutics. Moreover, RRM1 is located on chromosome 11p15.5, a region showing frequent loss of heterozygosity in NSCLC [9–11]. Additional earlier work suggests that over-expression of RRM1 contributes to enhanced resistance of cancer cells to gemcitabine [12]. Based on this combined information, we postulated that RRM1 could serve as a critical predictive biomarker for individualized chemotherapy of advanced NSCLC.

Several randomized, controlled trials assessing the role of RRM1 in predicting the efficacy of gemcitabine have been completed, but the findings are still inconsistent [13–16]. To further clarify the predictive value of RRM1 to chemotherapy, we performed a meta-analysis to evaluate the relevance of the expression of RRM1 to the clinical outcome of gemcitabine-based chemotherapy in advanced NSCLC.

Materials and methods

Literature search strategy

To identify studies involving associations between RRM1 expression levels and the clinical outcome of NSCLC before September 31, 2014, we conducted a literature search among three English databases including PubMed, Embase and Web of Science, and two Chinese databases including Wanfang and Chinese National Knowledge Infrastructure. We searched these electronic databases using the search terms “RRM1” and “lung” and “gemcitabine” to identify potential studies for inclusion in the current analysis. Additionally, we performed a computerized search of abstracts presented at the Annual Meetings of the American Society of Clinical Oncology (ASCO). Finally, we screened the references in all reviewed articles to identify additional articles that were not identified through the literature search described above.

Selection criteria

In our meta-analysis we included publications from studies meeting the following criteria: a) the patients had a pathological diagnosis of advanced NSCLC; b) all patients received gemcitabine-based chemotherapy for at least two cycles; c) the results were stratified by RRM1 expression level and RRM1 expression was detected by immunohistochemistry (IHC) or real-time reverse transcriptase PCR (RT-PCR); d) the results were part of an original analysis; e) the manuscripts were published in English or Chinese.

Data retrieval

Information from each study was retrieved independently by two investigators using a standardized data extraction form. Any discrepancy was resolved unanimously via discussion. The following information was recorded for each publication: first author’s name, publication date, study design, country of study, total number of patients included in the study, number of patients with RRM1 “high/positive” expression, number of patients with RRM1 “low/negative” expression, method used to ascertain RRM1 expression level, clinical stage of disease, detection method of RRM1 and Eastern Cooperative Oncology Group (ECOG) performance status. In cases where key information was lacking, the publication’s corresponding author was contacted. In the event that the given information could not be made available, it was classified as “not reported”.

Quality assessment

The quality and risk of bias within the publications were critically appraised separately by two reviewers. Quality assessment was conducted in each of the available studies by using the validated Newcastle-Ottawa Quality Assessment Scale for cohort studies [17,18]. This scale is composed of eight items that assess patient selection, study comparability and outcome with scores ranging from 0 to 9. Among our selections, the studies with a score of 6 or higher were graded as high quality.

Statistical methods

The endpoints in the pooled analyses were response rate and overall survival (OS) rate. RECIST criteria were utilized to define response, in which “complete response” or “partial response” were classified as “response”, and “stable” or “progressive” disease were classified as “non-response”. The odds ratio (OR) was abstracted or calculated to quantitatively evaluate the association between RRM1 expression level and response rate. The association between RRM1 level and OS was evaluated using the OR and its 95 % confidence interval (CI).

The pooled OR estimates were initially calculated using a fixed effects model. If the fixed effects P value for the I ² statistic was <0.10, which indicates significant heterogeneity across studies, the pooled estimate was further calculated using a random effects model. Additionally, in cases where there was qualitative evidence of methodological heterogeneity across studies (e.g., different RRM1 expression ascertainment methods), a random effects model was used.

In the subgroup analysis of RRM1 expression ascertainment methods, studies were classified as either using RT-PCR or IHC, as reported in the given publication. In the subgroup analysis of patient population type, studies conducted in China were classified as “Asian population”, whereas studies performed in other countries were classified as “European population”. All statistical analyses were carried out using RevMan 5.2 software.

Results and discussion

We performed a meta-analysis to evaluate the prognostic value of RRM1 expression levels and clinical outcome of gemcitabine-based chemotherapy in patients with advanced non-small cell lung cancer (NSCLC). The search terms “RRM1”, “lung” and “gemcitabine” yielded 510 relevant references in electronic databases. After exclusion of the studies that did not meet the selection criteria, we ended up with 12 studies encompassing 593 patients with advanced NSCLC. The flow chart of this search strategy is depicted in Fig. 1. All 12 selected studies were included in the pooled analysis of associations between RRM1 expression level and response rate, and three studies were included in the analysis of association between RRM1 expression level and overall survival (OS). A summary of the studies included in this meta-analysis is provided in Table 1. A quality assessment by the Newcastle-Ottawa Quality Assessment Scale for cohort studies revealed that the combined scores of selection, comparability and outcome were higher than 6 in each of the 12 selected studies (Table 2).

Fig. 1.

Electronic search flow chart

Table 1.

Summary of studies included in the meta-analysis

Author (ref)	No. patients per arm	Population	RRM1 measurement	Clinical stage	PS score	RRM1 Low/negative		RRM1 High/positive
						Total of cases	Overall response (No.)	Total of cases	Overall response (No.)
Bepler et al.[25]	35	European	RT-PCR	IIIA-IIIB	0–1	19	57.9 % (11)	16	25.0 % (4)
Boukovinas et al.[26]	95	European	RT-PCR	IIIB-IV	0–2	64	32.8 % (21)	31	25.8 % (8)
Gao et al.[27]	75	Asian	IHC	IIIB-IV	0–1	46	41.3 % (19)	29	31.1 % (9)
Jiang et al.[28]	60	Asian	IHC	IIIB-IV	0–2	26	57.7 % (15)	34	29.4 % (10)
Lee et al.[13]	38	Asian	IHC	IIIB-IV	0–3	25	24.0 % (6)	13	7.7 % (1)
Li et al.[29]	71	Asian	IHC	IIIB-IV	0–2	31	35.5 % (11)	40	7.5 % (3)
Liang et al.[30]	44	Asian	IHC	IIIB-IV	0–1	26	46.2 % (12)	18	27.8 % (5)
Liu et al.[31]	61	Asian	IHC	IIIB-IV	0–2	35	35.7 % (16)	26	19.2 %(5)
Rosell et al.[14]	16	European	RT-PCR	IIIB-IV	NR	12	50.0 % (6)	4	25.0 % (1)
Souglakos et al.[15]	42	European	RT-PCR	IIIB-IV	0–2	17	41.0 % (7)	25	24.0 % (6)
Su et al.[16]	22	Asian	RT-PCR	IIIB-IV	0–2	18	44.4 % (8)	4	0.0 % (0)
Wang et al.[32]	34	Asian	RT-PCR	IIIB-IV	0–2	17	52.9 % (9)	17	5.9 % (1)

Table 2.

Quality of literature included in the meta-analysis

Author	Year	Selection (score)	Comparability (score)	Outcome (score)
Bepler	2006	4	2	3
Boukovinas	2008	4	2	3
Gao	2011	4	2	3
Jiang	2013	4	2	3
Lee	2010	4	2	3
Li	2010	3	2	3
Liang	2012	3	2	3
Liu	2009	3	2	3
Rosell	2003	3	2	2
Souglakos	2008	4	2	3
Su	2010	4	2	3
Wang	2010	3	2	2

The quality of an included study is represented by the total points. Aggregate scores greater than 6 points mean high quality

The included studies used either retrospective or prospective observational designs. We did not find any clinical or methodological heterogeneity among the included studies, and the corresponding funnel plots were symmetrical (data not shown). Therefore, it is unlikely that publication bias had a major influence on the analyses performed.

The results of the pooled analysis of the association between RRM1 expression level and patient’s response rate are provided in Fig. 2. Since there was no heterogeneity among the studies (χ ² = 8.79, P = 0.64, I ² = 0 %), the fixed effects model was applied to perform the meta-analysis. As shown in Fig. 2, the number of patients with a high/positive RRM1 expression was 257, with an amalgamated effective rate of 20.6 %, whereas the number of patients with low/negative RRM1 expression was 336, with an amalgamated effective rate of 41.8 %. The response rate of patients with a low/negative RRM1 expression was significantly higher than that of patients with a high/positive RRM1 expression (OR: 0.35, 95 % CI: 0.24–0.51). We further applied the fixed effects model to pool the RT-PCR (χ ² = 5.42, P = 0.37, I ² = 8 %) and IHC (χ ² = 3.33, P = 0.65, I ² = 0 %) subgroup meta-analyses and found that the response rate of patients with low/negative RRM1 expression levels was significantly higher than that of patients with high/positive RRM1 expression levels, assessed in both IHC (41.8 % vs 20.6 %, OR = 0.34, 95 % CI 0.21–0.57, P < 0.00001) and RT-PCR (42.8 % vs 20.6 %, OR = 0.36, 95 % CI 0.20–0.65, P = 0.0008) subgroups (Fig. 3). A further population subgroup analysis with fixed effects model showed that there was no heterogeneity both in the Asian (χ ² = 6.14, P = 0.52, I ² = 0 %) and the European (χ ² = 1.59, P = 0.66, I ² = 0 %) subgroups (Fig. 3). Moreover, pooled meta-analysis of the subgroups revealed that the response rate of patients with low/negative RRM1 expression levels was also significantly higher than that in patients with high/positive RRM1 expression levels, both in Asian (42.9 % vs 18.8 %, OR = 0.30, 95 % CI 0.19–0.48, P < 0.00001) and European (40.2 % vs 25 %, OR = 0.48, 95 % CI 0.25–0.93, P = 0.03) subgroups. These results indicate that patients with low/negative RRM1 expression levels exhibit a better response to gemcitabine-containing chemotherapy, and that there is no significant heterogeneity among the outcomes obtained by RRM1 assessment method or patient population ethnicity. These analyses indicate that the RRM1 expression level is indeed a determinant of the response rate to gemcitabine-containing chemotherapy in advanced NSCLC patients.

Fig. 2.

Pooled analysis of RRM1 expression levels and response rates

Fig. 3.

Response rate to gemicitabine-containing regimens: subgroup analysis by detection method and ethnicity

We next assessed the one-year OS rates in three studies, which encompassed 196 patients. We performed a meta-analysis using the fixed effects model and found no heterogeneity among the studies (χ ² = 1.34, P = 0.51, I ² = 0 %). Importantly, we found that patients with low/negative RRM1 expression levels exhibited improved survival rates compared to patients with high/positive RRM1 expression levels (59.8 % vs 41.6 %, OR = 0.41, 95 % CI0.23–0.75, P = 0.004) (Fig. 4). These results indicate a statistically significant difference in OS between high and low RRM1 expressing NSCLC patients, and suggest that low/negative RRM1 expression is associated with a better OS of these patients.

Fig. 4.

Pooled analysis of RRM1 expression and one-year OS

So far, the most reliable routinely used methodologies to determine a biomarker status are IHC and PT-PCR. IHC detects gene expression at the protein level, whereas RT-PCR detects gene expression at the mRNA level. Zheng et al. [19] used these two methods simultaneously and found that the mRNA expression levels were closely related to the protein expression levels. We found that both the RT-PCR and IHC cohorts indicated that low/negative RRM1 expression corresponds to a better response to gemcitabine-based chemotherapy. However, IHC (OR = 0.34) seemed to be of greater value than RT-PCR (OR = 0.36). RT-PCR is currently the most sensitive method to detect expression while the IHC method is often more robust. Thus, the optimal method for evaluating RRM1 expression has yet to be defined and more prospective studies are needed to evaluate the efficacy of the RT-PCR method in detecting RRM1 expression levels.

We did not find any discrepancy in the subgroup analysis of the different populations. It thus remains to be established whether other factors such as pathological type, clinical stage of disease or ECOG performance status, have an impact on the RRM1-based survival outcomes. Our study was not designed to evaluate these relationships. Moreover, additional assessments of the association between RRM1 status and OS in European populations are needed.

The results of the present study are consistent with those of Gong et al. [20]. Compared to this latter study, however, we improved the methodology and included the most recently published data. First, the study by Gong et al. used median OS ratios in the pooled analysis of OS and time to progression, which have been shown to be suboptimal by Michiels et al. [21]. Secondly, the study by Gong et al. did not include recently published work. These two caveats were overcome by (i) assessing Odds Ratio’s to evaluate the association between RRM1 status and OS and (ii) including recently published studies (up to September 2014).

Apart from RRM1, other molecular markers such as ERCC1, BRCA1 and thymidylate synthase (TS) have been used to predict the response to chemotherapy in NSCLC patients. ERCC1 has been widely used for predicting the sensitivity of platinum-based chemotherapy of NSCLC [22,23]. A study by Qin et al. [22] also showed that high expression of BRCA1 was correlated with a shorter OS and progression free survival (PFS) compared to low expression. Lee et al. [24] found that low TS expression was significantly associated with a better response rate (RR) (P = 0.037) and a longer PFS (P < 0.001) in patients with pulmonary adenocarcinoma who were treated with pemetrexed/cisplatin as first-line chemotherapy. Therefore, more prospective studies with combined detection of these markers should be conducted to select the best for patients with NSCLC.

There are several limitations to the present study. Firstly, our conclusions are mainly based on findings from observational studies, which may potentially contain higher confounding factors than randomized controlled trials. Secondly, even though our funnel plot showed no evident publication bias among the included studies, our analysis was based on published studies. Since positive results are published more readily than negative results, publication bias remains an issue. Thirdly, the majority of studies included in the pooled analyses were on Asian populations, and it remains to be established whether the conclusions are equally applicable to other populations. Furthermore, our survey lacks enough individual patient data. Therefore, more details and subgroup data such as pathological types are needed in future analyses.

In conclusion, our findings support the hypothesis that RRM1 expression is associated with the response and overall survival rates of patients with advanced NSCLC who are treated with gemcitabine-based regimens. Our data suggest that RRM1 may serve as a predictive biomarker for individualized chemotherapy in advanced NSCLC. Large-scale prospective and randomized trials are required before RRM1 testing can be widely used as a prognostic tool in clinical settings.

Abbreviations

NSCLC

non-small cell lung cancer

RNR

ribonucleotide reductase

dNTPs

deoxyribonucleotides

ASCO

American Society of Clinical Oncology

IHC

immunohistochemistry

RT-PCR

real-time reverse transcriptase

OR

odds ratio

OS

overall survival

CI

confidence interval