Risk of Hyponatraemia in Cancer Patients Treated with Targeted Therapies: A Systematic Review and Meta-Analysis of Clinical Trials

Background Hyponatraemia has been reported with targeted therapies in cancer patients. Aim of the study was to perform an up-to-date meta-analysis in order to determine the incidence and relative risk (RR) in cancer patients treated with these agents. Materials and Methods The scientific literature regarding hyponatraemia was extensively reviewed using MEDLINE, PubMed, Embase and Cochrane databases. Eligible studies were selected according to PRISMA statement. Summary incidence, RR, and 95% Confidence Intervals were calculated using random-effects or fixed-effects models based on the heterogeneity of selected studies. Results 4803 potentially relevant trials were identified: of them, 13 randomized phase III studies were included in this meta-analysis. 6670 patients treated with 8 targeted agents were included: 2574 patients had hepatocellular carcinoma, whilst 4096 had other malignancies. The highest incidences of all-grade hyponatraemia were observed with the combination of brivanib and cetuximab (63.4) and pazopanib (31.7), while the lowest incidence was reported by afatinib (1.7). The highest incidence of high-grade hyponatraemia was reported by cetuximab (34.8), while the lowest incidences were reported by gefitinib (1.0). Summary RR of developing all-grade and high-grade hyponatraemia with targeted agents was 1.36 and 1.52, respectively. The highest RRs of all-grade and high-grade hyponatraemia were associated with brivanib (6.5 and 5.2, respectively). Grouping by drug category, the RR of high-grade hyponatraemia with angiogenesis inhibitors was 2.69 compared to anti-Epidermal Growth Factor Receptors agents (1.12). Conclusion Treatment with biological therapy in cancer patients is associated with a significant increased risk of hyponatraemia, therefore frequent clinical monitoring should be emphasized when managing targeted agents.


Introduction
Targeted therapies interfere with specific molecules involved in cancer cell growth, angiogenesis and survival, in contrast with traditional chemotherapy, drugs that mainly act against all actively dividing cells.
Such a different mechanism of action explains the absence of adverse events traditionally observed with cytotoxic chemotherapy and the occurrence of new drug-related toxicity profiles.
Among serum electrolytes disorders, hyponatraemia is probably the most frequent biochemical alterations potentially related to the use of these new agents.
Although many cases are asymptomatic, hyponatraemia may cause neurological symptoms, particularly when serum sodium declines rapidly or by a substantial extent [1].
Furthermore literature data suggest that hyponatraemia can be considered an unfavourable prognostic factor in this setting and it has been also hypothesized to adversely affect the response to anticancer treatment [2,3].
Moreover an effective and timely normalization of sodium levels could lead to a positive effect on prognosis of cancer patients.
The objective of the present study was to thoroughly assess incidence and relative risk of hyponatraemia in patients with solid tumors receiving targeted therapies through a revised meta-analysis of clinical trial in literature.

Selection of Studies
This systematic review and meta-analysis was achieved adhering to PRISMA guidelines for clinical trial selection [4]. PubMed and MEDLINE (since January 1966), Embase (since 1974) and the Cochrane Central Register of Controlled Trials (since 1967) quotations were revised in order to individuate studies of interest. In particular we selected more interesting trials generated from the research finding in Pubmed.
Full articles were obtained, and we checked for additional appropriate references. Where results were reported or updated in two or more publication, we selected the most recent or most thorough.
The primary objective of our study was to assess whether there is a correlation between hyponatraemia and treatment with targeted therapy.

Data Extraction and Quality Assessment
Data extraction was performed from full texts of eligible articles, by two independent evaluators (MS and EN). Data collected included author name list, year of publication, number of participants, treatment arms characteristic and targeted agent employed, number and grade (all-grade and high grade) of hyponatraemia cases reported in every arm.
National Cancer Institute's Common Terminology Criteria for Adverse Events (CTCAE) version 2 or 3 were applied to define adverse events (AEs). Study quality and appropriateness of randomization, double-blinding, and withdrawals was determined basing on Jadad scoring system [17].

Statistical Analysis
We considered the following summary measurements: incidence, relative risk (RR), and their corresponding 95% confidence intervals (CIs). Incidence evaluation was performed extracting from the safety section of eligible studies the following data: (1) number of patients receiving targeted therapy and (2) number of hyponatraemia cases. RRs of hyponatraemia was analyzed basing on data extracted from comparative trials in which patients were randomly assigned to receive targeted therapy or controls.
Cochran's Q test was applied to all variables to detect statistical heterogeneity among study outcomes; inconsistency of effects was measured using the I2 index as a parameter of inconsistency across studies attributable to heterogeneity and chance. Homogeneity of variance was violated for p values <0.1. Basing on Cochran's Q statistic significance it was decided whether to use random effects model (in case of significant Q test) or fixed effects model (in case of not significant Q test).
For each variable, model estimate and null hypothesis of overall non-significant difference between study-and control-group were tested. Microsoft Excel 2010 was employed to collect data; data analysis has been carried out with the "MATLAB and Statistics Toolbox Release 2012b".

Search Results
Four thousand, eight hundreds and three clinical trials studying target treatments employ in neoplastic patients resulted hypothetically significant to our research; of those, 2914 studies did not meet inclusion criteria due to any of the following causes: duplicate, phase I trials, not focused on targeted agents, reviews, observational studies, meta-analyses, case reports, letters or commentaries.

Quality of studies
The quality of trials selected for the meta-analysis was determined according to the Jadad assessment scale [31]. Follow-up time was appropriate in each trial. All the studies used either Common Terminology Criteria Adverse Events (CTCAE) version 2.0 or 3.0. Jadad scores of the 13 studies included are reported in Table 1. The average of Jadad scores was 3.7 (range 3 to 5) standing for an adequate quality of the meta-analysis.
An Eastern Cooperative Oncology Group performance status (ECOG-PS) not higher than 2 was required for enrolment in all 13 studies along with fair renal and hepatic functions, coagulation and haematological parameters. Baseline features of included trials are listed in Table 1. Selected target agents and their mechanism of action are listed in Table 2.
Three hundred and fifty six high-grade hyponatraemia occurrences were reported in patients assigned to treatment arms and 187 in the controls. The highest and lowest incidences of high-grade hyponatraemia were observed with cetuximab [22] (44.8%, 95% CI 32.0 to 57.6), and gefitinib [26] (1.0%, 95% CI 0 to 2.3), respectively. The incidences of all-grade and highgrade hyponatraemia are reported in Table 3.
In the overall study population, RR of all-grade and high-grade hyponatraemia was 1.

Discussion
Hyponatraemia represents an increasingly important issue in oncology clinical practice since it negatively correlates with performance status and with prognosis of cancer patients [32]. Patients with hyponatraemia have a higher risk of mortality and present a longer time of hospitalization with consequent cost increases [33]. An early detection and a prompt treatment of this disorder could prevent serious neurologic complication and improve overall survival (OS) [34]. For this reasons, it is pivotal for both physicians and patients to be aware about the risk of drug-induced hyponatraemia so as to promptly take the appropriate measures to face these events.
Our results show that the highest RR of all-grade hyponatraemia was associated with brivanib (RR = 6.5), whilst the highest RRs of high-grade hyponatraemia were reported by brivanib (RR = 5.2), sorafenib (RR = 2.4) and vorinostat (RR = 2.1). Moreover, grouping the selected agents into drug categories even strengthens these data. Indeed, the RR of developing high-  Recent studies demonstrated the activity of vascular endothelial growth factor (VEGF) in renal sodium metabolism, thus suggesting an activity of anti-VEGF/VEGF receptor agents in the homeostasis of sodium. Gu et al. evaluated the correlation between VEGF inhibition and hypertension. They found that rats receiving semaxanib (SU5416), a small-molecule inhibiting VEGF downstream signaling, showed increased mean arterial pressure and natriuresis. They also described a right shift with a slightly higher intercept of the pressure-natriuresis curve in rats with dietary salt-induced hypertension [35]. In addition, Grisk and colleagues reported that early hypertension induced by anti-VEGFR-TKI sunitinib is probably related to the direct action of this agent on the collecting ducts, suggesting a role for VEGFR-TKI in regulating renal sodium reabsorption [36].
Although the exact mechanisms underlying the increased incidence of hyponatraemia in patients exposed to targeted agents is still unclear, these evidences suggest an important role of VEGF/VEGFR pathway in sodium homeostasis. Most of the studies included in this analysis concerned hepatocellular carcinoma. In this condition, risk of developing hyponatremia is increased due also to concomitant liver cirrhosis that stimulates arginine vasopressin (AVP) secretion [37].
This study has several limitations. First of all it is a meta-analysis achieved starting from clinical trials and not from individual patients' data. This implies also the potential presence of confounding factors that were not considered, for instance patient comorbidities, previous administration of cytotoxic chemotherapy, and simultaneous treatments. In particular, in some of aforementioned trials, target agents were administrated with cisplatin. In this respect, literature data reported evidences about hyponatremia due to cisplatin-based chemotherapy, suggesting two possible mechanisms: stimulation of hypothalamic AVP production and damage of renal tubules with development of salt wasting syndrome [1,38]. Available data are insufficient to exclude a synergic effect between chemotherapy and target agents.
Furthermore, it should be consider that patients eligible for clinical trials mostly show fair organ functions, for this reason the incidence and severity of hyponatraemia may appear underrated in our meta-analysis comparing to clinical practice. Another limitation is the lack of data regarding delays, interruptions and discontinuations because of hyponatraemia, to correlate to our results.
In face of the limitations described, this meta-analysis, for the first time in literature, pointed out a correlation between targeted agents, in particular anti-angiogenetic ones, and hyponatraemia of all-and high-grade in patients with solid tumors. Moreover this study was the first to analyze RR of hyponatraemia in different groups of targeted agents, showing that maximum incidence of hyponatraemia was observed in patients treated with anti-VEGFR agents. Considering the negative prognostic and predictive role of hyponatraemia in cancer patients, a careful and prompt recognition of this event is preferred so as to limit negative consequences on patient outcome and to prevent possible treatment delays or interruptions.
Physicians and patients should be informed of such risks and an appropriate laboratory monitoring should be suggested to early detect hyponatraemia and optimize the management of these agents.

Conclusion
Hyponatraemia represents a negative prognostic factor for cancer patients. Increasing evidences showed a significant increased risk of developing hyponatraemia in patients treated with biological therapy. Therefore an accurate and frequent monitoring of serum sodium should be evaluated in patients treated with new-targeted agents, in particular with antiangiogenetic drugs, both in clinical practise and in prospective studies, for a rapid diagnosis and correction of this electrolyte disturbance.