Incidence and Factors Associated with Postoperative Delayed Hyponatremia after Transsphenoidal Pituitary Surgery: A Meta-Analysis and Systematic Review

Introduction Postoperative delayed hyponatremia is a complication associated with transsphenoidal pituitary surgery. Due to a wide spectrum of symptoms, the reported incidence and predictors of postoperative delayed hyponatremia vary among studies, and this deserves to be reviewed systematically. Methods PubMed, EMBASE, and CENTRAL databases were searched until September 1, 2020. Studies were included when (1) the event number of delayed hyponatremia after transsphenoidal pituitary surgery was reported, or (2) the associated factors of such complication were evaluated. Results A total of 27 studies were included for meta-analysis. The pooled incidence of overall and symptomatic delayed hyponatremia was 10.5% (95% confidence interval (CI) = 7.4–14.7%) and 5.0% (95% CI = 3.6–6.9%), respectively. No overt variations of the pooled estimates were observed upon subgroups stratified by endoscopic and microscopic procedure, publication year, and patients' age. In addition, 44.3% (95% CI = 29.6–60.2%) of unplanned hospital readmissions within 30 days were caused by delayed hyponatremia. Among the predictors evaluated, older age was the only significant factor associated with increased delayed hyponatremia (odds ratio = 1.16, 95% CI = 1.04–1.29, P = 0.006). Conclusion This meta-analysis and systematic review evaluated the incidence of postoperative delayed hyponatremia and found it as a major cause of unplanned readmissions after transsphenoidal pituitary surgery. Older patients are more prone to such complications and should be carefully followed. The retrospective nature and heterogeneity among the included studies and the small number of studies used for risk factor evaluation might weaken the corresponding results. Future prospective clinical studies are required to compensate for these limitations.


Introduction
Postoperative delayed hyponatremia, defined as serum sodium <130-135 mmol/L, can occur in 1.8-35% of patients who underwent transsphenoidal pituitary tumour surgery, with the peak incidence reported on day 7-8 following surgery [1][2][3]. Although the majority of patients with delayed hyponatremia are asymptomatic and recover spontaneously [4], patients whose serum sodium levels continue to drop below 125 mmol/L may become symptomatic [5]. Patients with hyponatremia may experience symptoms including headache, nausea, vomiting, lethargy, and confusion and can suffer from seizures, coma, and even death if left untreated in severe cases [1,2,6]. In particular, a serum sodium level of lower than 120 mmol/L was shown to associate with a 59.7% 5-year mortality [7]. Postoperative delayed hyponatremia was reported to be a common cause for unexpected hospital readmissions after transsphenoidal pituitary tumour surgery [2,8], and thus preventive measures have been implemented to reduce these events [9][10][11]. As such, identifying patients with increased risk of developing delayed hyponatremia is critical to the postoperative care of transsphenoidal pituitary surgery [12][13][14].
Several pathophysiological mechanisms of posttranssphenoidal pituitary surgery delayed hyponatremia have been proposed, with the syndrome of inappropriate antidiuretic hormone (SIADH) and cerebral salt wasting (CSW) syndrome as the main stream models [1][2][3]. Partial and complete disruption in the hypothalamic-pituitary gland regulation after surgery can lead to an overproduction of antidiuretic hormone and hence hyponatremia observed in SIADH. On the other hand, perturbed circulating natriuretic peptide levels and decreased sympathetic stimulation to the kidney associated with CSW are rare but potential causes of postoperative delayed hyponatremia in patients undergoing transsphenoidal pituitary surgery [1][2][3]. Association of patient sex, age, tumour size, and type of surgery with postoperative delayed hyponatremia has been illustrated previously [1][2][3]. Due to the spectrum of symptoms, disperse awareness of the complication, different aetiology, and the postoperative management protocols, the reported incidence and severity of delayed hyponatremia after transsphenoidal pituitary surgery vary widely and deserve to be scrutinized systematically. Furthermore, a detailed review on the risk factors for postoperative delayed hyponatremia is also warranted. e aims of this systematic review and meta-analysis are to address (1) the incidence of postoperative delayed hyponatremia following transsphenoidal pituitary surgery and its role on unplanned readmission and (2) the potential predictors of such complication by reviewing the published literature.

Search Strategy and Selection Criteria.
A literature search through the major public databases (i.e., PubMed, EMBASE, and CENTRAL) was conducted using "transsphenoidal," "pituitary neoplasms," "pituitary adenoma," "surgery," and "hyponatremia" as keywords combined with Boolean operators and using Medical Subject-Headings (MeSH) terms where appropriate for studies published prior to September 1, 2020. As an example, the search string used for PubMed was ("pituitary neoplasms/surgery" [MeSH Major Topic] OR "transsphenoidal" [All Fields]) AND hyponatremia [MeSH]. e inclusion criteria were studies that either reported (1) event number and/or incidence of delayed hyponatremia during the postoperative follow-up of either endoscopic or microscopic transsphenoidal surgery for pituitary tumours or (2) the associated factors of such delayed hyponatremia. In contrast, studies where (1) early or delayed onset hyponatremias cannot be clearly distinguished, (2) paediatric patients were enrolled only, (3) no quantitative outcomes of interest were available, and (4) language other than English was adopted were excluded from the present review. Regarding publication type, only cohort studies performed prospectively or retrospectively were eligible for inclusion, while letters, commentaries, editorials, proceedings, case reports, and personal communications were not considered. In addition, the reference lists of included studies were hand-searched to identify other potentially relevant studies.
is systematic review and meta-analysis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [15].

e Outcomes of Interest Included
(1) e incidence of postoperative delayed hyponatremia, defined as serum sodium concentration <130-135 mmol/L occurring on a postoperative day (POD) three and onwards (2) e incidence of symptomatic delayed hyponatremia, defined as delayed hyponatremia with patient-reported symptoms (3) e rate of postoperative delayed hyponatremia in unplanned hospital readmission within 30 days after surgery (4) e demographic and/or clinical factors associated with the occurrence of delayed hyponatremias e eligibility of studies identified via the above search and selection strategy was confirmed by two independent reviewers (CCL and YCH), and a third reviewer (CCC) was consulted where there was uncertainty regarding eligibility. From these eligible studies and when available, the following information was extracted: the name of the first author, year of publication, study design, study country, the total number of patients, type of surgery, patients' mean age, sex, tumour characteristics (i.e., functioning or nonfunctioning; macroor microadenomas), aetiology of hyponatremia, and main outcomes of interest. Due to the nature of the present review, raw patient data and private information were neither required nor used, and thus informed consent from study subjects or approval from the institutional review board was waived.

Quality Assessment.
We assessed the quality of included studies using the Newcastle-Ottawa scale (NOS) for cohort studies as recommended by the Cochrane Nonrandomized Studies Methods Working Group [16]. is scale awards a maximum of nine points to each study representing four points for the adequate selection of cohort participants, two points for the comparability of cohort participants on the basis of the design and analysis, and three points for the adequate ascertainment of outcomes. Quality assessment was performed by two independent reviewers (YCW and YTL), and a third reviewer (PWH) was consulted if any uncertainties occurred.

Statistical
Analysis. Event rates of the outcomes of interest with 95% confidence interval (CI) were extracted or calculated from each individual study, and then the summary effect with 95% CI was estimated. Planned subgroup analyses of overall postoperative delayed hyponatremia incidence were performed based on the type of surgery, publication year, and patients' mean age. For the associated factors, the OR with 95% CI was extracted from the included studies, and pooled estimates were generated.
An χ 2 -based test of homogeneity was performed to determine the inconsistency index (I 2 ) and Q statistics. If the I 2 statistic was >50%, the random-effects model was utilized to calculate pooled effects, and a fixed-effect model was employed when otherwise. A 2-sided P value of <0.05 was considered to indicate the statistical significance of the pooled estimate. Sensitivity analysis for the endpoints was addressed by the leave-one-out approach. Potential publication bias was assessed by Egger's test (whenever the total number of evaluated studies was >10) [17], respectively. e absence of publication bias was indicated by the data points forming a symmetric funnel-shaped distribution and a corresponding one-tailed P > 0.05. All analyses were performed using the Comprehensive Meta-Analysis statistical software, version 2.0 (Biostat, Englewood, NJ, USA).

Characteristics of Included Studies and Basic Patient
Demographics.
In general, the definition used for postoperative delayed hyponatremia was comparable among studies, with serum sodium of <130-135 mmol/L as the cut-off and onset on or beyond postoperative day 3 (Table 2). e incidence of overall delayed hyponatremia and those with symptoms in the individual study was in the range of 2-35% and 2.1-19.8%, respectively. Among the 6 studies that reported delayed hyponatremia as a cause of unplanned hospital readmission within 30 days after the initial transsphenoidal surgery, the rate of delayed hyponatremia ranged from 21% to 61% [11,20,27,29,30,36].

Sensitivity Analysis and Publication Bias.
Sensitivity analyses were performed using the leave-one-out approach in which meta-analysis was performed with each study removed in turn (Table 4). e summary effects of overall incidence of postoperative delayed hyponatremia, symptomatic delayed hyponatremia, and readmission due to delayed hyponatremia remained between 10 and 11%, unchanged (5%), and 41-49%, respectively, indicating that the meta-analysis had good reliability and the data was not overly influenced by each study. Figure 3 shows that there was no publication bias in the findings with regard to the incidence of overall delayed hyponatremia (t � 1.106, P � 0.140, Figure 3(a)) and symptomatic delayed hyponatremia (t � 0.153, P � 0.440, Figure 3(b)) as exemplified by Egger's test. e rate of readmissions resulting from postoperative delayed hyponatremia was not subjected to Egger's test due to the limited number of studies included (i.e., 6 in total).

Quality
Assessment. e quality rating of the individual study is shown in Table 1. e total score ranged from 6 to 8 with an average of 6.3, suggesting that the studies included were of moderate quality (Table 1).

Discussion
e present study is the most updated systematic review and the first in reporting the results of a corresponding metaanalysis on the incidence and risk factors of postoperative delayed hyponatremia after transsphenoidal pituitary surgery. We found that the pooled incidences of postoperative delayed hyponatremia overall and those with symptoms were 10 and 5%, respectively. Delayed hyponatremia contributed to 44% of the unplanned hospital readmissions within 30 days. Significant heterogeneity existed among the studies, whereas these estimates were robust under sensitivity analyses. Among different subgroups categorized by surgical type, mean age, and publication year, the incidences seemed not to vary greatly from the main findings. Older age was associated with a slightly increased risk of postoperative delayed hyponatremia, while male gender, tumour type, and size showed no significant associations.
Previously, Cote et al. (2016) have systematically reviewed the incidence and predictors of delayed hyponatremia after transsphenoidal pituitary surgery based on 10 studies with a total of 2,947 patients [14] which were all included in our meta-analysis. e authors have focused on symptomatic delayed hyponatremia only and reported an event rate of about 4-12%, with a variety of potential predictors proposed. However, no quantitative estimations were conducted. An American registry-based study surveyed the postoperative complications in 1,240 patients who underwent transsphenoidal pituitary surgery during 2006-2015 and found hyponatremia as the major cause for unplanned rehospitalizations, representing 29.5% of such readmissions [8].
Among the studies evaluating predictors of delayed hyponatremia after transsphenoidal pituitary surgery, most studies found that older age was associated with the occurrence of delayed hyponatremia [12,13,23,[26][27][28]. Despite the numerous reports testing the predictive values of females, macroadenoma, and a diagnosis of Cushing's disease in delayed hyponatremia [4,6,12,13,22,[24][25][26][27][28][30][31][32], the result of the present meta-analysis indicated that older age was the only significant associated factor. e studies that observed significant associations between sex, tumour size, or tumour pathology and postoperative delayed hyponatremia were mostly published prior to 2013 [4,22,24,26,31], where potential differences in the awareness of the complication and postoperative management approach from the study afterwards may exist and thus led to inconsistency. SIADH, which commonly results from iatrogenic injury of the neurohypophysis during surgical exploration and leads to degeneration of magnocellular neurons and excessive release of antidiuretic hormone, is proposed as an aetiology of postoperative delayed hyponatremia [1,2]. Another less commonly accepted explanation for such complication is CWS secondary to increased release of atrial and cerebral natriuretic peptide. Other potential causes of postoperative serum sodium perturbation include adrenocortical insufficiency, hypothyroidism, hypernatremia overcorrection, and volume overload. Although our original aim was to evaluate the pooled rates of delayed hyponatremia due to different aetiologies, this is precluded by the limited studies that reported definite aetiologies in the literature.
To date, there are only limited studies that directly compared the potential impact of endoscopic versus microscopic transsphenoidal pituitary surgery on the occurrence of delayed hyponatremia. It may be intuitive to speculate that less iatrogenic injury in the neurohypophysis is associated with endoscopic than microscopic procedures due to improved visualization. However, a prior systematic review compared postoperative complications between these two surgical approaches and only found 3 published articles reporting relevant data [38], with no significant differences for the SIADH between endoscopic and microscopic procedures. e present and a separate systematic review [14] arrived at similar conclusions in that the postoperative delayed hyponatremia rate did not differ with regard to endoscopic versus microscopic procedures. Future welldesigned randomized controlled trials are warranted to confirm this finding.
Due to the increased awareness of postoperative delayed hyponatremia and the evidence showing such complication as the main cause of unplanned readmissions after transsphenoidal pituitary surgery, clinicians have devoted to constructing and implementing outpatient care algorithm in preventing delayed hyponatremia-associated rehospitalizations [12][13][14]. In brief, the preventive measures involved a fluid restriction (to 1.0-1.5 L/day) that was mandatory or guided by an active screening of serum sodium levels 5-7 days after discharge from the hospital. Based on these reports, mandatory fluid restriction starting from the day of discharge from the hospital was vital in preventing the occurrence of symptomatic hyponatremia that resulted in readmission [10,11], whereas active screening-guided fluid intake control was not sufficient in this respect [9]. ese findings perhaps reflect the fact that SIADH is the more prevalent aetiology of delayed hyponatremia after transsphenoidal pituitary surgery than CSW, where hyper-and hypovolemia contribute to the pathophysiology of the two respective conditions [2].
As with previously published studies of similar nature, the interpretation of the results of the present meta-analysis is limited by the heterogeneity across the eligible studies, although they were selected based on stringent criteria and a thorough literature search. To address this issue, we have adopted the random-effects model, study stratification, and sensitivity analysis. Nonetheless, no specific source of heterogeneity could be indicated. Additionally, the retrospective nature of most included studies presents as a potential source of selection bias overall, whereas the small number of studies used for risk factor evaluation may weaken the corresponding results. In the present study, all functional adenomas were pooled together, and we were not able to look at Cushing's disease and acromegaly separately because most of the included studies did not separately report the outcomes by tumour pathologies. Only a rough categorization of age as a risk factor (above 50 vs. below 50) was done in the present analysis due to the nature of data extracted. More research efforts and prospective clinical studies in the field are required to compensate for the limitations discussed herein, especially with an in-depth investigation on the risk factor age.

Conclusion
Taking together, this systematic review and meta-analysis demonstrated an incidence of hyponatremia which is unignorable following transsphenoidal pituitary surgery. Furthermore, delayed hyponatremia is a major cause of 30day readmissions in these patients. Among the potential risk factors, only old age is associated with this complication. Despite several limitations, the present findings may enhance the surgeons' awareness of the patient profile with an increased risk of delayed hyponatremia. Based on such understanding, clinicians may be motivated to develop preventive or interventive measures to reduce the possibility and impact of postoperative delayed hyponatremias.
Data Availability e datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Risk factor
Odds ratio Odds ratio and 95% CI Figure 4: Meta-analysis for the associations between delayed hyponatremia and age, gender, tumour type, and tumour size.
International Journal of Endocrinology 11