4.1. Etiology of Post-transsphenoidal Hyponatremia
Disorders of sodium balance are common after transsphenoidal surgery [19
]. The mechanism underlying this predisposition is not entirely clear, but some authors have suggested that it may be attributable to the aberrant release of hormones (e.g., ADH from the neurohypophysis, cortisol from the adenohypophysis) or abnormal sympathetic hypothalamic outflow occurring as a result of surgical manipulation of the gland itself, the stalk or the hypothalamus [19
] and others [32
] described a now classic, triphasic trend of serum sodium abnormalities seen after transection of the infundibular stalk in both animals and humans (in which DI is initially seen, followed by hyponatremia related to SIADH several days later and a final phase consisting of DI, which often persists). Indeed, much of the early literature suggests that post-transsphenoidal hyponatremia (particularly when arising in a stereotypically delayed fashion, 5–7 days after surgery) is predominantly due to SIADH [33
]. The classic triphasic response described by Fisher is only seen in 1%–2% of patients undergoing transsphenoidal surgery (1.6% in our series), however [19
], and many patients with post-transsphenoidal hyponatremia demonstrate features (e.g., volume depletion) altogether inconsistent with SIADH. This discrepancy between clinical and experimental observation may be a manifestation of interventions (e.g., fluid restriction) administered in the clinical setting (but not in the experimental setting), as suggested by some authors [19
], or alternatively, an illustration of this experimental model’s insufficiency in accurately characterizing the underlying pathology of post-transsphenoidal hyponatremia.
Olson et al.
prospectively studied 92 patients undergoing transsphenoidal surgery, 25% of whom developed postoperative hyponatremia. Evidence of impaired suppression of ADH-secretion was seen not only in the hyponatremic patients, but also in two-thirds of the normonatremic patients. They demonstrated that the patients with impaired ADH suppression who failed to develop hyponatremia exhibited higher dietary sodium intake and lower natriuresis compared with hyponatremic patients, suggesting that the pathophysiology of post-transsphenoidal hyponatremia is complex and multifactorial and not due only to impaired ADH suppression, but instead displaying features of both SIADH and CSW [21
Several reports of post-transsphenoidal hyponatremia in the literature suggest CSW as a primary etiology [38
]. The pathophysiology of CSW is poorly understood, but CSW after transsphenoidal surgery has been suggested to arise as a result of the aberrant release of brain natriuretic peptide (BNP) and other natriuretic factors to the manipulation of the hypothalamic-pituitary axis, demonstrated in some studies to be related to an underlying abnormality of sympathetic outflow from the hypothalamus to the renal system, resulting in decreased renal tubular sodium resorption [40
]. BNP [42
], atrial natriuretic peptide (ANP) [44
] and other natriuretic factors [46
] have been suggested to serve as the molecular mediators of this abnormal natriuresis in CSW, although conclusive evidence of this relationship is lacking, and no identifiable consensus currently exists [4
Secondary hypoadrenalism must also be considered in the differential diagnosis of post-transsphenoidal hyponatremia. Some authors suggest that the characteristically-delayed hyponatremia arising after transsphenoidal surgery is most often due to adrenocortical insufficiency [47
]. As cortisol is known to inhibit vasopressin secretion [49
], a role for hypoadrenalism in producing the SIADH-like effect seen in many patients with delayed post-transsphenoidal hyponatremia is not difficult to imagine.
In our series, while the majority of post-operative hyponatremia was felt to be related to SIADH (71%), characteristics suggestive of CSW (e.g., hypovolemia, natriuresis) were seen in 24.2%. Secondary adrenal insufficiency was considered a contributing factor in 6.5%, over-administration of desmopressin acetate in 4.8% and excessive oral intake of hypotonic fluids in 1.6%.
Interestingly, patients in this series subjected to post-operative CSF diversion were at a significantly greater risk of hyponatremia development when compared with patients in whom CSF drainage did not occur (p
= 0.0006), although neither the length (R
= 0.023, p
= 0.91) of CSF drainage nor the rate (R
= 0.11, p
= 0.59) of CSF drainage correlated with the severity of hyponatremia in these patients. Although the development of hyponatremia in direct relationship to CSF drainage has been reported only rarely in several case reports and small series [51
], water retention in association with extrarenal sodium losses are a well-documented etiology of hypotonic, hypovolemic hyponatremia (e.g., as seen with excessive gastrointestinal losses) [55
], and a similar mechanism is likely the underlying cause of hyponatremia development in these patients.
4.2. Incidence of Post-Transsphenoidal Hyponatremia
Though the reported incidences of post-transsphenoidal hypernatremia and hyponatremia vary considerably, results of most large, modern series indicate that hyponatremia is more common [17
]. The incidence of post-transsphenoidal hyponatremia in this series (18%) is consistent with that seen in other large series (16%–38.8%) [19
]. It has been suggested that the relatively stringent monitoring of electrolyte values seen in this series led to the early identification and treatment of hyponatremia, which may have otherwise developed into a delayed, more severe hyponatremia, thus falsely minimizing the incidence of delayed hyponatremia in this series. In this regard, it should be made clear that the majority of patients (17/27 patients, 63.0%) with early-onset (post-operative Day 0 or post-operative Day 1 onset) post-transsphenoidal hyponatremia in our series did not receive any treatment for their hyponatremia, and thus, it is felt unlikely that early treatment in these cases masked a delayed, more severe and/or more symptomatic hyponatremia. Furthermore, although data regarding the length of ICU stay and the stringency of laboratory monitoring trends after transsphenoidal surgery is relatively lacking in the literature, many large series report laboratory monitoring trends consistent with, or with greater stringency than, those described in this series [37
Hyponatremia has been shown to lead to a considerable degree of morbidity and mortality in the literature [1
]. Although no complications or persistent clinical consequences occurred in our series as a direct result of hyponatremia, the length of ICU stay (OR
= 1.43, p
≤ 0.0005) and hospital stay (OR
= 2.82, p
< 0.0005) were both significantly higher in patients with hyponatremia.
In this series, a significantly higher incidence of post-operative hyponatremia was seen in female patients (p
= 0.011, OR
= 2.18) and in patients with cardiac, renal or thyroid disease (p
= 0.0034, OR
= 2.60), even when these disease were well-controlled. Other studies have shown a similar predisposition toward hyponatremia in female patients. The mechanism underlying this gender predilection is not entirely clear, but previous studies have documented a gender-specific difference in the antidiuretic response to antidiuretic hormone [59
], a finding that may be particularly relevant after transsphenoidal surgery, as ADH suppression has been shown to be impaired in many post-transsphenoidal surgical patients [19
]. Several prior series have documented an elevated risk of post-transsphenoidal hyponatremia in patients with macroadenomas [34
], while others have shown an elevated risk for hyponatremia in microadenomas (proposed by some authors to be related to a greater degree of intraoperative exploration and hypophyseal manipulation occurring during microadenoma resection) [21
]. This relationship between tumor size and hyponatremia risk was not seen in our series. Similarly, a large series by Jahangiri et al.
documented an elevated risk for post-transsphenoidal hyponatremia in patients with preoperative hypopituitarism [56
], but these findings were not reproduced in our series.
4.3. Onset and Clinical Manifestations of Post-Transsphenoidal Hyponatremia
According to the classically-described triphasic pattern of sodium imbalance after infundibular transection, polyuria and hypernatremia (due to diabetes insipidus) occur initially (2–3 days after transection) and are followed several days later (days 5–7 after transection) by oliguria and hyponatremia [31
]. This timeline of events has also been observed clinically in some patients after transsphenoidal surgery and other neurosurgical procedures in or around the hypothalamic-hypophyseal system [36
]. This so-called “delayed postoperative hyponatremia” is thought to arise due to the slow and delayed release of ADH stored within the magnocellular osmoregulatory neurons of the posterior lobe of the pituitary gland prior to the injury event [21
], a theory bolstered in part by the experimental finding that this delayed hyponatremia can be avoided by preemptive removal of the neurohypophysis in canines [64
The results of our study highlight an important caveat to the above principle, however, which is that the majority of hyponatremia (when defined as a serum sodium <135 mEq/L) actually occurs very early in the post-operative course (mean onset in our series: 3.2 ± 4.6 days postoperatively), is quantitatively mild, often asymptomatic and tends to resolve promptly and spontaneously. Though the mechanism underlying this association between early hyponatremia and mild hyponatremia is not entirely clear, some authors propose that this mild, early post-transsphenoidal hyponatremia is actually a different entity altogether and not related to ADH release, but rather to excessive administration of hypotonic fluids during surgery and in the immediate postoperative period [37
]. The spontaneous resolution of this early hyponatremia could thus be explained as being due to renal clearance of this excess fluid volume, a function that occurs relatively rapidly in patients without renal disease.
That the majority of patients with post-transsphenoidal hyponatremia lack identifiable clinical symptoms relatable to hyponatremia has been documented previously [19
]. The clinical manifestations of hyponatremia often correlate well with the severity of hyponatremia (where mild hyponatremia is more often asymptomatic), as seen in this series.
4.4. Treatment of Post-Transsphenoidal Hyponatremia
The treatment for post-transsphenoidal hyponatremia should ideally be tailored to the underlying etiology in each case. The treatment for SIADH consists primarily of fluid restriction, as the prevailing pathophysiology in SIADH lies in inappropriate resorption of free water in the setting of normo or euvolemia, as a result of the influence of ADH on the late distal tubules and collecting ducts of the nephron [2
]. The “gold standard” treatment for CSW, in contrast, is fluid resuscitation, although the most appropriate fluid type for replacement in CSW is a matter of debate [2
]. Normal saline has traditionally been used, although early studies indicate that hypertonic saline is both safe [68
] and effective [2
Exogenous mineralocorticoids have also been used to treat hyponatremia in neurosurgical patients with some success, both in the setting of CSW [69
] and of hyponatremia related to secondary adrenal insufficiency [2
]. Early series examining the use of vasopressin receptor antagonists (e.g., conivaptan or tolvaptan) have also shown promising results, and though costly, these agents may be particularly useful in acute, severe hyponatremia, such as that seen in patients initially discharged and later readmitted with severe, symptomatic hyponatremia [56
The use of hypertonic saline (OR = −2.4), sodium chloride tablets (OR = −1.57) and normal saline (OR = −0.44) were the most effective treatments in our series, regardless of hyponatremia etiology or severity, although the predictive value of these factors did not achieve statistical significance. Curiously, the use of fluid restriction tended to lengthen the required treatment in our series, even when used in patients with suspected SIADH.
4.5. Study Weaknesses
The evaluation of hyponatremia in the neurosurgical patient, as in any patient or setting, should ideally progress systematically and according to a series of characteristic laboratory tests and clinical evaluations, all of which have been well-defined elsewhere [24
]. In the clinical setting, however, certain steps in this evaluation can be overlooked and certain assumptions made, making the retrospective evaluation of hyponatremia etiology challenging. In several instances in our series, for example, a low serum osmolality was documented by laboratory analysis. In most cases, hyponatremia was presumed to be of the hypotonic variety, as this is the most common scenario, and no evidence existed to suggest otherwise (e.g., no patient was severely hyperglycemic, no patient was given mannitol and no history of administration of sodium-free irrigant solutes was present). Similarly, 24.2% of the hyponatremia in our series was felt to be related to CSW, although in most of these cases, urine sodium was not measured, and thus, the possibility that the hypovolemic hyponatremia observed in some of these patients was related instead to extrarenal solute loss (e.g., GI losses, skin losses) cannot be reliably ruled out. CSF diversion via a lumbar drain (seen in 22.7% of all patients, 40.3% of patients with hyponatremia of any etiology and 73.3% of hypovolemic hyponatremia, felt to be due to CSW) was likely to be a contributing factor, or perhaps the predominant underlying etiology, in a considerable portion of the patients with post-transsphenoidal hypovolemic hyponatremia in this series. If CSF diversion were theoretically considered to be the primary etiology of hyponatremia development in the 11 patients in this series with hypovolemic hyponatremia attributed to CSW, but in whom CSF diversion also occurred, then the incidence of hyponatremia due to CSW would decrease to 6.4% (from 24.2% as reported), and the incidence of hyponatremia due to extrarenal sodium loss would be 17.7%.
Furthermore, volume status, which plays a prominent role in the evaluation of hyponatremia, was assessed in this series by physical exam findings and other non-invasive measures. Assessing volume status by non-invasive measures, however, has been shown to be relatively unreliable, and doing so retrospectively may be particularly problematic. Neither volume status prior to surgery nor uric acid levels at the time of hyponatremia were routinely documented in this series. Additionally, although volume status was systematically monitored (through non-invasive monitoring of fluid intake and output) intraoperatively and post-operatively for patients in the ICU setting, volume status assessments were likely less reliable once patients were transferred to the regular, acute care setting.
Other weaknesses of this study include the tendency to use of a variety of different treatment methodologies for hyponatremia in the same patient (e.g., fluid restriction, hypertonic saline and NaCl tablets together), a practice that may not be unusual in the clinical setting, but one that makes it challenging to accurately evaluate the effectiveness of each treatment strategy in isolation. Furthermore, vasopressin receptor antagonists, which have been shown in some series to have a significantly greater efficacy in the treatment of post-transsphenoidal hyponatremia when compared with conventional therapies, were utilized in only a single patient in this series, preventing one from drawing reliable conclusions regarding their effectiveness in this patient population.
Somewhat surprising to the authors was the fact that no single treatment strategy used in this series was shown to significantly improve the rate of hyponatremia correction. In fact, many of the treatment strategies were found to negatively impact the rate of hyponatremia correction. This finding may be at least partially explained as a manifestation of selection bias, in that patients with more severe hyponatremia were more likely to receive one or more treatments for hyponatremia (compared to those with mild hyponatremia, who were more likely to go untreated), and thus, these treatment strategies appear to be associated with more severe hyponatremia and a slower rate of correction, whereas the converse statement may in fact be true. A randomized, prospective trial would be required to provide definitive evidence regarding the effectiveness of these and other treatment strategies for post-transsphenoidal hyponatremia.