Prognostic role of lactate dehydrogenase in solid tumors: A systematic review and meta-analysis of 76 studies

Abstract

Background. In cancer cells, metabolism is shifted to aerobic glycolysis with lactate production coupled with a higher uptake of glucose as the main energy source. Lactate dehydrogenase (LDH) catalyzes the reduction of pyruvate to form lactate, and serum level is often raised in aggressive cancer and hematological malignancies. We have assessed the prognostic value of LDH in solid tumors.

Material and methods. A systematic review of electronic databases was conducted to identify publications exploring the association of LDH with clinical outcome in solid tumors. Overall survival (OS) was the primary outcome, and cancer-specific survival (CSS), progression-free survival (PFS), and disease-free survival (DFS) were secondary outcomes. Data from studies reporting a hazard ratio (HR) and 95% confidence interval (CI) were pooled in a meta-analysis. Pooled HRs were computed and weighted using generic inverse-variance and random-effect modeling. All statistical tests were two-sided.

Results. Seventy-six studies comprising 22 882 patients, mainly with advanced disease, were included in the analysis. Median cut-off of serum LDH was 245 U/L. Overall, higher LDH levels were associated with a HR for OS of 1.7 (95% CI 1.62–1.79; p < 0.00001) in 73 studies. The prognostic effect was highest in renal cell, melanoma, gastric, prostate, nasopharyngeal and lung cancers (all p < 0.00001). HRs for PFS was 1.75 (all p < 0.0001).

Conclusions. A high serum LDH level is associated with a poor survival in solid tumors, in particular melanoma, prostate and renal cell carcinomas, and can be used as a useful and inexpensive prognostic biomarker in metastatic carcinomas.

The different metabolism of cancer compared with that of normal cells confers a selective advantage for their proliferation and survival. The use of lactate as an energy source requires the conversion of lactate into pyruvate as well as the transport of lactate into and out of tumor cells by specific transporters. The former pathway is regulated by lactate dehydrogenase (LDH). LDHs catalyze the conversion of pyruvate and lactate, with concomitant conversion of NADH and NAD+. These enzymes are homo- or hetero-tetramers composed of a M and a H protein subunits that are encoded by LDHA and LDHB genes, respectively. LDHA and pyruvate dehydrogenase kinase (PDK) are up-regulated in solid tumors in response to hypoxia in a HIF-1α-dependent manner. LDHA reducing the pyruvate into lactate regenerates the NAD+ pool necessary to maintain an adequate glycolytic process. The combinations of LDHA and LDHB proteins into tetramers result in the five major isoforms of LDH, numbered 1–5. LDH-5 has the highest efficiency to catalyze the conversion of pyruvate to lactate. LDH-5 is mainly localized to the cytoplasm, where it participates in glucose metabolism. Expression of LDHA and increased levels of functionally active LDH-5 are commonly observed events in highly invasive hypoxic cancers that are resistant to chemo- and radiotherapy. Biochemical markers of tumor burden, such as LDH have been incorporated into several prognostic scores and staging for renal cell carcinoma, melanoma and colorectal cancer [1–3]. In metastatic germ cell tumors, furthermore, LDH is incorporated in the IGCCCG prognostic classification, and this possibly aids in guiding treatment and stratifying patients in clinical trials [4].

The significance and magnitude of the prognostic impact of circulating LDH are, however, unclear. LDH might be a simple and inexpensive objective prognostic parameter that could be used in daily oncologic clinical practice, other than helping to stratify patients in clinical trials. The aim of this study was to summarize all data to quantify the prognostic value of LDH on the clinical outcome in various solid tumors, using standard meta-analysis techniques.

Material and methods

This analysis was conducted in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines [5].

Search methods and criteria for selecting studies for this review

An electronic search of PubMed, EMBASE, SCOPUS, Web of Science, CINAHL and the Coch- rane Register of Controlled Trials was performed. Search terms included: (“neoplasms”[MeSH Terms] OR “neoplasms”[All Fields] OR “cancer”[All Fields]) AND (LDH[All Fields] OR (“l-lactate dehydro- genase”[MeSH Terms] OR (“l-lactate”[All Fields] AND “dehydrogenase”[All Fields]) OR “l-lactate dehydrogenase”[All Fields] OR (“lactate”[All Fields] AND “dehydrogenase”[All Fields]) OR “lactate dehydrogenase”[All Fields])) AND (“mortality” [Subheading] OR “mortality”[All Fields] OR “survival”[All Fields] OR “survival”[MeSH Terms]) AND (multivariate[All Fields] OR (cox[All Fields] AND (“regression (psychology)”[MeSH Terms] OR (“regression”[All Fields] AND “(psychology)”[All Fields]) OR “regression (psychology)”[All Fields] OR “regression”[All Fields]))) AND (hazard[All Fields] AND (“Ratio (Oxf)”[Journal] OR “ratio”[All Fields])). Citation lists of retrieved articles were screened manually to ensure the sensitivity of the search strategy.

Inclusion criteria for the primary analysis were as follows: 1) studies published in complete papers and in the English language, of (at least 10) adult patients with solid tumors reporting on the prognostic impact of serum LDH; and 2) availability of a hazard ratio (HR) and 95% confidence interval (CI) for overall survival (OS). For a secondary analysis, studies providing an HR for cancer-specific survival (CSS), progression-free survival (PFS) or time to progression, and disease-free survival (DFS), were included as well. Duplicate publications were excluded. Two reviewers (F. Petrelli, M. Cabiddu) evaluated independently all of the titles identified by the search strategy. The results were then pooled, and all potentially relevant publications were retrieved in full. The same two reviewers then evaluated the complete articles for eligibility. To avoid inclusion of duplicated or overlapping data, we compared author names and institutions where patients were recruited and if substantial doubts remained, the study reporting fewer patients was not included in the analysis.

Data extraction

OS was the primary outcome of interest. CSS, PFS, and DFS were secondary outcomes. The following details were extracted: name of first author, type of study, year of publication, number of patients included in the analysis, disease site, disease stage (non-metastatic, metastatic, mixed [non-metastatic and metastatic]), cut-off defining high LDH, and HRs for OS, PFS, DFS, or CSS as applicable. HRs were extracted only from multivariable analyses. When two or more HRs were reported for different cut-off values, only those above the median LDH value for that study were considered.

Data collection and statistical analysis

The meta-analysis was conducted initially for all included studies for each of the endpoints of interest. Subgroup analyses were conducted for disease site and stage of disease only for the primary outcome. Extracted data were aggregated into a meta-analysis using RevMan 5.3 software (Cochrane Collaboration, Copenhagen, Denmark). Estimates of HRs were weighted and pooled using the generic inverse-variance and random-effect model [6]. Analyses were conducted for all studies, and differences between diseases were assessed using methods described by Deeks et al. [7]. Meta-regression analysis was performed to evaluate the effect of LDH cut-off level (when available) on the HR for OS. Publication bias for the primary endpoint was assessed by visual inspection of the funnel plot. Heterogeneity was assessed with the use of the Cochran Q and I² statistics. All statistical tests were two-sided, and statistical significance was defined as p < 0.05.

Results

Seventy-six studies [8–83] with a total of 22 882 patients were included (Figure 1). Characteristics of the included studies are shown in (Supplementary Table I available online at http://informahealthcare.com/doi/abs/10.3109/0284186X.2015.1043026); most (95%) were published in the last decade. Among trials, eight studies regarded gastrointestinal malignancies (4 biliopancreatic, 2 gastric, 1 colorectal and 1 neuroendocrine cancers), seven head and neck cancers, seven lung cancers, 12 melanomas, one sarcoma, one thymic carcinoma, one mesothelioma, one breast cancer, four mixed or unknown primaries, 16 prostate cancers, 15 renal cell carcinomas and three testicular cancers. Number of patients ranged from 35 to 2425. In all trials patients with mainly metastatic or locally advanced unresectable disease were included (only 5 trials included stage I–III cancers).

Figure 1. Study flow diagram of included studies.

Overall survival

Seventy-three studies reported HR for OS. Forty-one studies did not specify a numerical cut-off used for multivariate analysis; conversely, in 34 studies a predefined cut-off was reported (median: 245 U/L). Twelve of the eligible 73 studies (16%) reported a non-statistically significant HR.

Overall, LDH level was associated with a HR for OS of 1.7 (95% CI 1.62–1.79; p < 0.00001; I^{2 =} 92%, p for heterogeneity < 0.00001, random-effect model). The effect of LDH level on OS among disease subgroups is shown in Figure 2. The prognostic effect of serum LDH, among subgroups with two or more studies, was highest in renal cell carcinoma (HR = 2.08; 95% CI 1.74–2.5, p < 0.00001), followed by melanoma (HR = 1.97; 95% CI 1.59–2.43, p < 0.00001), gastric cancer (HR = 1.91; 95% CI 1.38–2.63, p < 0.0001), prostate cancer (HR = 1.9; 95% CI 1.58–2.29, p < 0.00001), nasopharyngeal carcinoma (HR = 1.74; 95% CI 1.52–1.99, p < 0.00001), and lung cancer (HR = 1.54; 95% CI 1.33–1.78, p < 0.00001). The HR for the subgroup of mixed solid tumors series or unknown primary was 1.74 (95% CI 1.4–2.15, p < 0.00001). Differences between disease subgroups were statistically significant (p for subgroup difference < 0.00001). Neuroendocrine tumors, breast cancer, thymic carcinoma, mesothelioma, testicular and colorectal cancer, and sarcoma data were provided only in one paper each.

Figure 2. Forest plots showing hazard ratio for overall survival for LDH greater than or less than the cut-off for overall studies and by disease subgroups.

For the 15 disease-site subgroups analyzed, there was statistically significant heterogeneity among trials of biliopancreatic cancer, renal cell carcinoma, prostate cancer, and melanoma, whereas heterogeneity among other trials was non-statistically significant. The scatter plot for the meta-regression is shown in (Supplementary Figure 1 available online at http://informahealthcare.com/doi/abs/10.3109/0284186X.2015.1043026). Overall, there was a nearly statistically significant association between LDH cut-off and the HR for OS (p = 0.11). There was evidence of significant publication bias, with several studies reporting significantly higher results than expected (Figure 3).

Figure 3. Funnel plot of hazard ratio for overall survival for serum LDH level (horizontal axis) and the standard error (SE) for the hazard ratio (vertical axis). Each study is represented by one circle. The vertical line represents the pooled effect estimate.

Cancer-specific survival

Few studies reported this data, so a formal meta-analysis was not performed.

Progression-free survival or time to progression

Thirteen trials reported HRs for PFS or TTP. Overall, a high LDH level was associated with a poor PFS (HR = 1.75, 95% CI 1.31–2.33, p < 0.0001; I² = 86%, p for heterogeneity < 0.0001, random-effect model).

Disease-free survival

Only two studies including nasopharyngeal cancer patients reported HRs for DFS. A high LDH concentration was associated with a lower DFS as multivariate analysis (HR = 1.68, 95% CI 1.24–2.27, p = 0.001; I² = 47%, p for heterogeneity = 0.16, fixed effect model).

Discussion

Several studies have suggested that elevated LDH is both a hallmark of aggressive carcinomas and lymphomas and associated with an unfavorable outcome. Indeed, LDH levels are actually included in the TNM staging system for melanoma [3], defining the M1c subgroup that is associated with a five-year OS of 10%. LDH is also considered with respect to the risk status of testicular cancer, in particular metastatic non-seminoma germ cell tumors [84].

We performed a meta-analysis of 76 studies involving 22 882 patients with solid tumors in order to assess the prognostic effect of serum LDH. We found a consistent impact of elevated LDH on survival (HR = 1.7) among various disease subgroups, and mainly in advanced disease stages. High serum LDH was also associated with a reduced PFS in 13 trials. As expected, the magnitude of the effect on OS was the greatest in melanoma and renal cell carcinoma other than metastatic (castration-resistant) prostate cancer. The burden of the data considered (about 50% of the included studies) concerns these neoplastic diseases, in which hypoxia and angiogenic factors play a crucial role. In particular, in renal cell carcinoma, a five-parameter-based prognostic score provided by the Memorial Sloan-Kettering Cancer Center that is commonly used in patients with metastatic disease includes LDH levels and correlates these with outcomes in patients treated in various clinical trials before and after the era of targeted therapy [85,86]. The present analysis mainly covered patients with advanced tumors, which confirms that the prognostic role of LDH currently seems to be confined to only highly proliferative and metastatic cancers. We focused on solid tumors, because the prognostic role of LDH in hematological disease is well known, especially in high-grade lymphoma where it is elevated in almost 50% of patients [87]. The novelty of our data is that it particularly relates to prostate cancer, where the role of LDH as a prognostic factor has not yet been fully demonstrated. Several papers reporting the role of inflammation parameters (neutrophil-lympocyte ratio and C-reactive protein) other than bone-related biomarkers (e.g. alkaline phosphatase) have recently been published [88–94]. In particular, LDH seems to be associated with a poor prognosis in metastatic, castration-resistant prostate cancer, and could thus be valuable for identifying patients who are suitable for the earlier use of chemotherapy. Data on LDH levels and inflammatory parameters could be useful when starting therapy for prostate cancer and the other metastatic malignancies examined in this meta-analysis. It is possible that C-reactive protein reflects the enhanced immune response elicited by a tumor, which is more intense when tissue necrosis (and so LDH) and inflammation are high, as occurs with more aggressive cancer subtypes.

That lactates play a role in cancer cells is unsurprising, and LDH concentrations in human serum during neoplastic conditions can be regarded as a marker of (cancer) the metabolic activity and avid glucose uptake of cells. Tumor hypoxia actually induces, among other things, the expression of a hypoxia-inducible factor (HIF), which is a transcription factor that initiates a range of activities, including angiogenesis and various prosurvival mechanisms [95]. In terms of tumor growth, the local blood supply becomes inadequate, thus requiring angiogenesis and leading to hypoxia. The metabolism of cells is consecutively shifted towards the anerobic glycolytic pathway (e.g. from glucose to lactate) through the enhanced expression of glycolytic enzymes and glucose transporters. This is coupled with a reduced dependence on the oxidative pathway. Cancers are also able to produce lactate due to increased glycolytic activity, regardless of oxygen availability (the so-called Warburg effect), which is a phenomenon that was first described about a century ago as being a consequence of energy requests from tumor cells. Lactate which is the end product of glycolysis from pyruvate, is produced in large amounts in highly aggressive tumors in response to the specific characteristics of the microenvironment and the genetic features of neoplastic cells. Importantly, lactate also constitutes an alternative metabolic source for cancer cells [96–98]. The LDH-A gene promoter possesses two conserved hypoxia response elements containing functionally-essential binding sites for HIF-1 with the consensus sequence 5′-RCGTG-3′, which may strongly suggest an oxygen-dependent regulation of LDH-5 activity [99].

Recently, high LDH levels have been found to predict the response to anti-angiogenetic agents, such as bevacizumab and sorafenib. In two papers, Scartozzi et al. examined this predictive role, demonstrating that: 1) in subjects with high serum LDH, bevacizumab confers a higher RR in colorectal cancer; and 2) in hepatocellular carcinoma treated with sorafenib, LDH levels seem to predict clinical outcomes in terms of PFS and OS [100,101].

This meta-analysis does, however, have some limitations. First, only published data, as opposed to individual patient data, have been collected. Second, we found evidence of high heterogeneity among the trials, as well as a significant publication bias, with a larger number of trials than expected reporting a significantly higher effect of LDH levels. Furthermore, we only included studies reporting HRs, while more than 20 publications reporting on the prognostic value of LDH were excluded (e.g. because only odds ratios, relative risk for survival, or p for significance were reported), possibly introducing bias. It is notable that almost all of the excluded papers (the majority analyzing genitourinary tumors, sarcomas, melanomas and lung cancers) presented a significant association with outcome, thus reducing the risk that could have changed the final results. Many other case series, however, were excluded because serum LDH data was unavailable in the population analyzed. Third, LDH is influenced by other non-neoplastic conditions, including heart failure, hypothyroidism, anemia, and lung or liver disease, which may have influenced the serum concentrations. Most of the studies included in this meta-analysis did not explicitly control for such different, benign conditions that may have affected LDH levels. Finally, a cut-off value associated with a significantly worse OS cannot be defined, even if the median of the reported cut-off values is the common upper normal limit used in most laboratories (245 U/L). Unfortunately, only about 50% of the papers considered provided details of such a cut-off point, and the metaregression for correlations of serum cut-off levels with HRs was not significant. However, to our knowledge, this is the first meta-analysis to systematically confirm the poor prognostic value of high LDH concentrations in solid tumors. It also involves 22 961 patients included in 76 published studies, mainly concerning genitourinary and melanoma tumors. The value of the analysis does, however, remain confined to advanced tumors with locally advanced and metastatic disease, and its significance with respect to the early stages of disease requires further evaluation.

Recent in vitro studies have shown that inhibiting LDHA expression may lessen invasiveness and the metastatic potential of cancer cells by reducing their proliferation capacity and reversing their resistance to chemotherapy [102–110]. LDH can thus be regarded as an ideal target for anticancer therapies.

In conclusion, serum LDH is associated with poor survival with respect to many solid tumors, and could be used as a cost-effective prognostic biomarker in terms of prognostic scores before treating advanced disease and enrolling patients in prospective trials. The evaluation of LDH as a therapeutic target is also warranted, and is the subject of ongoing trials.

Supplementary material available online

Supplementary Figure 1 and Table I available online at http://informahealthcare.com/doi/abs/10.3109/0284186X.2015.1043026.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.