Abstract
Background Pituitary apoplexy is often treated urgently, but this is based on studies which assess vision categorically and dichotomizes the time interval between symptom onset and surgery which may introduce bias in measuring an association between the two.
Objective This study was aimed to assess for a relationship between continuously valued surgery delay and visual acuity recovery after pituitary apoplexy.
Methods In this retrospective study, all patients presenting with symptomatic pituitary apoplexy between 2004 and 2016 were identified from an institutional database. The primary endpoint was visual acuity improvement, measured as the difference in acuity from the pre- to postoperative period, and measured in Logarithm of the Minimal Angle of Resolution (LogMAR) units. Analysis was performed using continuous values of time delay and visual acuity to assess for an underlying association.
Results Thirty-two pituitary apoplexy patients were identified. Visual acuity deficits were reported in 81%. The median visual acuity was 0.35 LogMAR (Snellen's fraction 20/40) preoperatively, and 0.1 (20/25) postoperatively ( p < 0.01). The time between symptom onset and surgery was not associated with improvement in visual acuity ( p = 0.46). When the time delay and visual outcome were intentionally dichotomized, patients undergoing surgery within 2 days of symptom onset had 0.21 times the odds of a good visual outcome (95% confidence interval [CI]: 0.04–1.05).
Conclusion When assessed as continuously valued measures, the time from symptom onset to surgical intervention and the improvement in visual acuity are not associated, although intentional dichotomization of data produced conflicting results.
Keywords: acuity, early, delayed pituitary apoplexy, timing, vision
Introduction
The prevalence of pituitary adenomas is estimated at 19 to 28 cases per 100,000 people. 1 Pituitary apoplexy—a rare phenomenon of hemorrhage and/or infarction of the gland or neoplastic tissue—affects 2 to 12% of patients with pituitary adenomas and 75 to 80% of these patients have no known history of adenoma at the time of presentation. 2 3 While the most common presenting symptom is headache, apoplexy may also cause vision loss, oculomotor deficits, adrenal insufficiency, and/or severe electrolyte disturbances, and may be life-threatening.
Pituitary apoplexy has historically been considered as a neurosurgical emergency which warrants expedient surgical treatment. Recent studies suggest a role for conservative management in some situations, though most patients with severe or worsening visual deficits, oculomotor palsies, or altered mental status still require surgical decompression. 4
The optimal timing or urgency of surgical decompression for those patients requiring intervention remains undetermined. 5 Numerous small, retrospective studies have assessed the relationship between early and late surgery and improvement in vision, though these rely on a variety of definitions of “early” and often measure visual outcome categorically as fully, partially, or not recovered. 5 6 Such use of dichotomous definitions of “early” surgery and categorical vision outcome measures may lead to spurious associations.
The aim of the current study was to assess the relationship between the time interval from symptom onset to surgical decompression and postoperative recovery in visual acuity. We hypothesized that when time delay and visual acuity are assessed as continuously valued measures, prior associations would no longer be significant.
Methods
This single-center, retrospective consecutive case series study was designed to assess the relationship between timing of surgery and visual or endocrinologic outcomes in pituitary apoplexy. Individual patient consent was not required, as the study was performed in a retrospective fashion and deidentified patient data were used for analysis. The study was approved by the local institutional review board.
Population
Using an institutional database of all adult patients undergoing transsphenoidal surgery for pituitary tumors from 2004 to 2016, we extracted all patients who had a clinical history consistent with symptomatic pituitary apoplexy (i.e., sudden onset of headache, visual symptoms, or cranial neuropathy) who presented to the emergency department for evaluation. Because the database included only surgical cases, patients undergoing medical management only were not included. Patients were excluded if there was no imaging, intraoperative, or histological evidence of hemorrhage or pituitary infarction. Imaging evidence of hemorrhage or infarction was determined by an attending neuroradiologist; presence of necrosis, hemorrhage, or infarcted tissue was confirmed via review of permanent section pathology reports.
Institutional Practice
Patients presenting to our institution's emergency department with suspected pituitary apoplexy undergo endocrinologic workup (including basic metabolic panel, serum and urine osmolality, random serum cortisol, adrenocorticotrophic hormone, prolactin, follicle stimulating hormone, luteinizing hormone, insulin-like growth factor-1, thyroid stimulating hormone, and free T4), visual assessment by an attending ophthalmologist and magnetic resonance imaging with and without contrast.
The neurosurgical management paradigm is to intervene urgently if there is an acute, severe visual field deficit, visual acuity decline, or cranial neuropathy, or if there is progressive worsening of symptoms. Patients with minimal or stable/improving deficits are often managed conservatively; this population is not assessed in this study. In some cases, surgery is performed prior to formal ophthalmological assessment if that assessment is delayed and is not expected to change management.
All patients receive stress-dose glucocorticoids preoperatively. Patients undergo endocrinologic laboratory assessment on the morning of postoperative day (POD) 1. Postoperative imaging is obtained during hospitalization only if clinically indicated; routine magnetic resonance imaging is obtained at outpatient follow-up. Patients are discharged home as early as POD 1, if the thirst mechanism is intact and laboratory workup does not suggest diabetes insipidus. All patients receive a 6-week course of glucocorticoid supplementation in keeping with our published institutional protocol. 7
Patients are scheduled for regular follow-up in a combined endocrinology/neurosurgery pituitary clinic at 6 weeks and 3 months after surgery. Patients undergo postoperative imaging and hypothalamic-pituitary axis assessment at the 6-week follow-up appointment. Additional follow-up appointments, laboratory testing, and imaging are scheduled based on individual patient needs.
Data Collection
Patient demographics, medical comorbidities, presenting symptoms, details of the surgical procedure, pre- and postoperative imaging characteristics, ophthalmologic assessments, endocrinology evaluations, and follow-up data were obtained by review of the electronic medical record. For follow-up ophthalmology and endocrinology assessments, data were obtained from the most recent available note.
The primary outcome measure of the study was visual acuity; endocrinologic outcomes were included as a secondary outcome measure. Hypopituitarism of any axis was defined by either signs or symptoms of endocrine dysfunction, laboratory testing, or persistent hormone replacement therapy.
The timing of onset of symptoms was determined by review of consultation or history and physical notes in the chart. As the exact time of symptom onset was often not available, the date of symptom onset was used. In instances where there was discrepancy among notes, the note which gave the most specific description of the timing of onset of symptoms was selected. When the date was specified only by a week or month, the earliest date of the specified time interval was used.
Visual acuity assessments given as Snellen's scores were converted to Logarithm of the Minimal Angle of Resolution (LogMAR) values for analysis (LogMAR = −log[Snellen fraction]). Conversion to LogMAR allows for analysis of visual acuity as a continuously-valued measure. Nonnumeric visual acuity measures including finger counting (FC), hand motion (HM), light perception (LP), and no light perception (NLP) were ascribed LogMAR values of 1.9, 2.3, 2.7, and 3.0, respectively, in keeping with previously published methods. 8
Statistical Analysis
Patients without follow-up vision or endocrinology assessments were included in the descriptive statistics but were not included in the final analysis. No data were imputed. Since visual acuity measures included approximate values for the nonnumeric measures and data were generally not normally distributed, nonparametric statistical tests (Fisher's exact and Wilcoxon's rank-sum tests, unless otherwise noted) were used. All analysis was performed in Stata/IC version 15.1. 9
Results
Patient Population
A total of 32 patients were identified who underwent transsphenoidal surgery for symptomatic pituitary apoplexy at our institution during the 2004 to 2016 period, with median endocrinology and ophthalmology follow-up of 259 days (interquartile range [IQR]: 43–1,005 days) and 49 days (IQR: 4–189 days), respectively.
All patients had complete pre- and postoperative endocrinology assessments. Formal visual acuity assessments were not available for one patient preoperatively; an additional four patients did not have formal postoperative acuity assessments, leaving 27 patients with complete pre- and postoperative visual acuity measures. Four patients with missing formal acuity measures did have subjective vision assessments, so a total of 31 patients had categorical vision outcomes (i.e., improved, stable, and worsened). The baseline demographic information and comorbidities for this study population are shown in Table 1 .
Table 1. Demographics and baseline characteristics of apoplexy population.
| Variable | Value ( n = 32) a | |
|---|---|---|
| Male gender | 18 | (56) |
| Age at presentation (y) | 58 | (42–63) |
| Height (cm) | 172 | (166–183) |
| Weight (kg) | 85 | (70–107) |
| BMI | 28 | (25–33) |
| Comorbidities | ||
| Known pituitary mass | 6 | (19) |
| OSA | 1 | (3) |
| Hypertension | 8 | (25) |
| IICH | 1 | (3) |
| Tumor characteristics | ||
| Hemorrhagic | 24 | (75) |
| Cystic | 7 | (22) |
| Largest dimension (cm) | 3.0 | (2.2–3.5) |
| Regional invasion | ||
| Suprasellar | 31 | (97) |
| Cavernous sinus | 14 | (44) |
| Histology | ||
| Nonfunctioning adenoma | 23 | (72) |
| Functioning adenoma | 9 | (28) |
Abbreviations: BMI, body mass index; IICH, idiopathic intracranial hypertension; OSA, obstructive sleep apnea.
Values represent number of patients (%) for categorical variables and median (interquartile range) for continuous variables.
The most frequent presenting symptom was headache ( Table 2 ). Eighty percent of patients reported subjective decreased visual acuity and 73% described a subjective visual field deficit. Bitemporal hemianopsia was specifically noted in 19% of patients. Thirteen patients had ocular motor palsies (cranial nerves III, IV, VI, or a combination thereof).
Table 2. Presenting symptoms.
| Variable | Value ( n = 32) a | |
|---|---|---|
| Headache | 26 | (81) |
| Nausea/vomiting | 15 | (44) |
| Photophobia | 3 | (9) |
| Fever | 0 | (0) |
| Altered mental status | 2 | (6) |
| Decreased visual acuity b | ||
| Bilateral | 11 | (35) |
| Unilateral | 14 | (45) |
| Neither | 6 | (19) |
| Visual field deficit c | ||
| Bilateral | 15 | (50) |
| Unilateral | 7 | (23) |
| Neither | 8 | (27) |
| CN-III neuropathy | ||
| Bilateral | 2 | (6) |
| Unilateral | 6 | (19) |
| Neither | 24 | (75) |
| CN-IV neuropathy | ||
| Bilateral | 0 | (0) |
| Unilateral | 4 | (12) |
| Neither | 28 | (88) |
| CN-VI neuropathy | ||
| Bilateral | 0 | (0) |
| Unilateral | 7 | (22) |
| Neither | 25 | (78) |
Abbreviation: CN, cranial nerve.
Values represent number of patients (%) for categorical variables, unless otherwise specified.
One patient was unable to endorse a visual acuity deficit due to altered mental status.
Two patients were unable to participate in a visual field exam due to altered mental status.
Timing of Symptom Onset, Presentation, and Surgery
The median time from onset of symptoms to presentation was 3 days (range: 0–14), whereas the median time between symptom onset and surgical intervention was 4 days (range: 1–15; Fig. 1 ). There was no association between preoperative acuity and time between symptom onset and presentation (linear regression, p = 0.82). All patients underwent decompression of the optic apparatus and tumor resection via a transsphenoidal approach within a median of 1 day of presentation to the hospital (range: 0–4, Fig. 1 ). Imaging and pathological characteristics of the tumors are listed in Table 1 .
Fig. 1.
Histograms of time intervals between symptom onset and surgery, with overlaid kernel density estimation plots. Top left: onset of symptoms to initial evaluation by neurosurgery; Bottom left: onset of symptoms to surgical intervention; Top right: Initial evaluation by neurosurgery to surgical intervention.
Ophthalmologic Outcomes
Pre- and postoperative visual acuity and endocrine assessments are listed in Table 3 . For one patient, assessment by the ophthalmologist was unable to be obtained prior to urgent surgical intervention. Four patients had missing postoperative visual acuity data; two followed-up with outside providers and records were unavailable and two declined ophthalmology follow-up.
Table 3. Preoperative and Postoperative ophthalmology and endocrinology assessments.
| Variable | Preoperative | Postoperative | p -Value | ||
|---|---|---|---|---|---|
| Visual acuity | |||||
| Right | 0.18 (20/30) | (0–1.3) | 0.1 (20/25) | (0–0.3) | 0.08 |
| Left | 0.7 (20/100) | (0–2.3) | 0.1 (20/25) | (0–0.6) | 0.02 |
| Worst eye | 1.3 (20/400) | (0.3–3) | 0.18 (20/30) | (0–2) | <0.01 |
| Endocrinologic function | |||||
| Adrenocortical insufficiency a | 11 | (34) | 10 | (31) | 1.0 |
| Hypothyroidism b | 14 | (44) | 18 | (56) | 0.34 |
| Hypogonadism c | 5 | (16) | 7 | (22) | 0.69 |
| Diabetes insipidus d | 2 | (6) | 5 | (16) | 0.38 |
| Other e | 4 | (13) | 5 | (16) | 1.0 |
Note: visual acuity shown in LogMAR units, with equivalent Snellen fractions in brackets. Continuous measures are shown as median (interquartile range) and were assessed with Wilcoxon's signed-rank tests. Binary measures are written as n (%) and were assessed with McNemar's exact test.
Determined by serum adrenocorticotrophic hormone, cortisol stimulation, or persistent glucocorticoid replacement.
Determined by serum thyroid stimulating hormone and free T4 or persistent thyroxine supplementation.
Determined by serum follicle stimulating hormone or luteinizing hormone or hormone replacement therapy.
Determined by serum or urine electrolytes when available or persistent desmopressin prescription.
Other category includes elevated prolactin, growth hormone deficiency/excess, syndrome of inappropriate antidiuretic hormone secretion (SIADH), unspecified hyponatremia, or otherwise unspecified pituitary dysfunction.
There was a statistically significant overall improvement in visual acuity after surgery (Wilcoxon's sign-rank test, p < 0.01). Median preoperative visual acuity was 0.35 LogMAR (20/40 Snellen's fraction, range 20/20—NLP), and median postoperative acuity was 0.1 LogMAR (20/25 Snellen's fraction, range 20/20—NLP). Pre- and postoperative visual acuity measures are plotted in Fig. 2 . Nine of 13 ocular motor palsies had resolved at last follow-up.
Fig. 2.
Right: Scatter plot of pre-operative and post-operative visual acuity. Left: Scatter plot of the change in visual acuity as a function of time between symptom onset and surgical intervention. A small amount of noise (“jitter”) is added to each point to avoid overlapping markers. FC, finger counting, HM, detects hand motion, LP, light perception, NLP, no light perception.
Endocrinologic Outcomes
There were no statistically significant differences between endocrinologic function before or after surgery for the total cohort. There was a higher rate of postoperative adrenocortical insufficiency and hypothyroidism among patients with functioning adenomas (67 and 89%, respectively) compared with nonfunctioning adenomas (17 and 43%), however.
Assessment of Timing of Surgery and Visual Outcomes
Univariate analyses with Wilcoxon's rank-sum or Kruskal–Wallis rank tests were performed using presenting/preoperative factors (chronological, clinical, or radiographic) which were suspected to associate with visual acuity recovery ( Table 4 ). The change in visual acuity was defined as the postoperative LogMAR acuity in the worst eye minus the preoperative acuity in that eye for each patient (i.e., a negative result indicates greater improvement in acuity). For each continuous variable, the relationship with visual acuity improvement was graphed to assess for any nonlinear relationships (not shown).
Table 4. Univariate analysis of suspected predictors of visual acuity improvement.
| Variable | n | Value ( n = 27) a | p -Value c | |
|---|---|---|---|---|
| Time from symptom onset to surgery (d) | 27 | −0.15 | (25) | 0.46 |
| Worst preoperative acuity measure | 27 | −0.22 | (25) | 0.28 |
| Age (y) | 27 | 0.01 | (25) | 0.95 |
| Maximum tumor dimension (cm) | 27 | −0.20 | (25) | 0.33 |
| Hemorrhage No Yes |
7 20 |
−0.06 −0.18 |
(−1.2 to 0.1) (−0.95 to 0) |
0.67 |
| Hypertension No Yes |
20 7 |
−0.18 0 |
(−0.95 to 0) (−1.2 to 0) |
0.76 |
| Suprasellar tumor extension No Yes |
1 26 |
−0.18 −0.12 |
(−1.0 to 0) | 0.95 |
| Visual acuity deficit None Unilateral Bilateral |
6 12 9 |
0 −0.3 −0.7 |
(−0.18 to 0.1) (−1.2 to 0) (−0.9 to 0) |
0.29 |
|
Visual field deficit
b
None Unilateral Bilateral |
7 7 12 |
0 0 −0.9 |
(−0.18 to 0.1) (−1.0 to 0) (−1.2 to −0.03) |
0.10 |
For continuous variables, Spearman's Rho and degrees of freedom are displayed as R s (df) with associated p -values. For binary variables, the median (interquartile range) of visual acuity change is shown for each value.
Visual fields were assessed via confrontation. Visual fields were not able to be obtained for one patient due to altered mental status.
Binary predictors were assessed using a Wilcoxon's rank-sum test; categorical predictors with a Kruskal–Wallis rank test.
There was no significant association between visual acuity improvement and time between symptom onset and surgery ( Fig. 2 ). None of the predictors assessed demonstrated a significant relationship with visual acuity outcome. Due to the small sample, this study was not appropriately powered to perform a multivariate analysis.
For comparison with prior studies, these univariate analyses were repeated using the categorical assessment of visual acuity change (i.e., improved vs. stable or worsened), rather than the continuously valued measure. Of the 31 patients with subjective vision assessments, 20 reported improvement in their vision after surgery and 7 reported that their vision was stable. None of the predictors had a statistically significant association with improved visual acuity, though presence of unilateral or bilateral subjective visual acuity deficits and field deficits showed a nonsignificant trend toward greater improvement ( Table 5 ).
Table 5. Univariate analysis of suspected predictors with dichotomous visual outcome.
| Variable | Visual acuity b | p -Value c | |||
|---|---|---|---|---|---|
| Improved ( n = 20) |
Stable/worse ( n = 11) |
||||
| Time from symptom onset to surgery (d) | 4.5 | (3–7.5) | 2 | (2–5) | 0.34 |
| Worst preoperative acuity measure (LogMAR) | 1 | (0.54–1.9) | 1.9 | (0.1–3) | 0.95 |
| Age (years) | 59 | (42–66) | 53 | (35–65) | 0.85 |
| Maximum tumor dimension (cm) | 3 | (2.4–3.5) | 2.5 | (1.9–3) | 0.23 |
| Hemorrhage No Yes |
5 15 |
(25) (75) |
3 8 |
(27) (73) |
1.0 |
| Hypertension No Yes |
17 3 |
(85) (15) |
7 4 |
(64) (36) |
0.21 |
| Suprasellar tumor extension No Yes |
1 19 |
(5) (95) |
0 11 |
(0) (100) |
0.65 |
|
Visual acuity deficit
b
None Unilateral Bilateral |
2 8 9 |
(11) (42) (47) |
4 5 2 |
(36) (45) (18) |
0.05 |
|
Visual field deficit
b
None Unilateral Bilateral |
4 3 12 |
(21) (16) (63) |
4 4 2 |
(4) (4) (2) |
0.06 |
For continuous variables, median (interquartile range) are shown. For binary variables the number of patients with visual improvement is displayed as n (%).
Visual fields were assessed via confrontation. Visual acuity and visual fields were not able to be obtained for one and two patients, respectively, due to altered mental status.
A Wilcoxon's rank-sum test was used for continuous variables, Fisher's exact test for dichotomous variables, and a χ 2 test for trend for ordinal variables.
To assess for the effect of dichotomizing, the time from symptom onset to surgery—as performed in other studies—the time variable was dichotomized at 1, 2, 3, 5, 7, 10, and 14 days, most of which have been previously assessed as a dichotomization time in other studies. 10 11 12 13 14 15 Odds ratios of improved vision based on each dichotomized value were determined by univariate logistic regression, as shown in Fig. 3 . In this analysis, patients undergoing surgery within 2 days of symptom onset had 0.21 times the odds of a good visual outcome (95% confidence interval [CI]: 0.04–1.05, p = 0.05). A univariate linear regression of visual acuity improvement was then performed using each dichotomized value of time delay as a predictor; none of these associations were statistically significant. Coefficients for each regression are also shown in Fig. 3 .
Fig. 3.
Regression analysis of surgical delay. The time cutoff is the dichotomization point of the delay from symptom onset to surgery (days). Univariate regressions were performed independently without a reference. Top: Univariate logistic regression analysis of time from symptom onset to surgery in predicting improved vision (versus stable or worsening vision). Greater odds ratio indicates greater probability of a good visual outcome. The time cutoff of 14 days was unable to be calculated as only one patient had time between symptom onset and surgery greater than 14 days. Bottom: Linear regression coefficients for univariate analysis of time from symptom onset to surgery in predicting change in visual acuity. A negative change in acuity indicates visual improvement in LogMAR units.
Discussion
This retrospective cohort study demonstrated that there was no association between the continuously valued measures of time from symptom onset to surgical intervention and postoperative recovery in visual acuity in patients presenting with pituitary apoplexy.
Prior Evidence
In 1993, Bills et al published a series of 18 patients with visual acuity deficits due to pituitary apoplexy which demonstrated a significantly higher rate of complete visual recovery when surgical decompression was performed within 7 days of symptom onset. 16 Since that seminal, highly cited study, numerous case series and cohort studies have demonstrated varied associations between the timing interval from symptom onset to surgical decompression and improvement in vision or oculomotor deficits and debate in the literature persists.
Many of these studies relied on categorical assessment of visual change after surgery (often based on neurosurgical vision assessments) and dichotomized the time interval between symptom onset and surgery. This may have represented the only available data for those studies, but reducing continuously valued data to categorical variables results in loss of a considerable amount of information, introduces bias, and increases residual confounding. 17 18 19 20 Furthermore, use of dichotomous outcome data increases the necessary sample size to perform multivariable logistic regression (vs. multivariable linear regression for continuous data). The purpose of this study was to assess what differences might arise when the continuously valued data are utilized.
Patient Characteristics
Our institutional cohort exhibited baseline characteristics consistent with the epidemiology of pituitary apoplexy, including a male predominance, median age in the fifth to sixth decade, previously undiagnosed tumors in the majority of patients, and headache as the most frequent presenting symptom. 21 Visual field deficits are found in 56 to 82% of patients with pituitary apoplexy, and the rate was 73% in this cohort. 6 Visual acuity deficit was common and although the median preoperative visual acuity was 20/40, the range of acuity was wide (20/20 to NLP). Patients presented within 2 weeks of symptom onset, and the clear majority underwent transsphenoidal surgery within 1 day of presentation.
Visual Acuity Recovery
In keeping with the findings of other studies, there was a statistically significant improvement in vision after surgery (median postoperative acuity of 20/25), including improvement for a portion of patients who presented with complete blindness. 22 23 Rates of various endocrinopathies were unchanged after surgery consistent with the finding that 73 to 96% of patients will have persistent postoperative pituitary dysfunction. 24
In a univariate analysis, the time from symptom onset to surgery was not found to be significantly associated with visual improvement. The Spearman's rho (R s ) of −0.15 for this association suggests that only approximately 2% of the variability in visual recovery is attributable to the time delay from symptom onset to surgery. Although limited by the univariate nature of this analysis, this suggests that earlier intervention may not lead to improved acuity outcomes. Similarly, Rutkowski et al found no relationship between time from symptom onset to surgery and a binary assessment of visual acuity improvement. 25
Few predictors of visual recovery in pituitary apoplexy have been published in the literature. Analysis of postoperative visual recovering in nonapoplectic pituitary adenomas has suggested that preoperative acuity level, preoperative visual field deficit, and duration of symptoms are associated with greater recovery, though only upper temporal field deficit severity was independently predictive of outcome. 26 Our cohort did demonstrate a nonsignificant trend toward greater acuity improvement when there were either subjective acuity or field deficits.
Unfortunately, formal perimetry assessments were not available for these patients owing to their urgent presentation. As described above, visual field deficit severity is believed to be associated with outcome, and field deficit may sometimes dictate whether to intervene and how urgently to do so. A quantitative visual field assessment may be more informative for predicting outcome but such formal assessments are not regularly performed in the emergency setting at our institution. Accordingly, only the presence of a gross field deficit on confrontation was included in this analysis. Without formal visual field data, we were not able to account for these confounding effects or the interdependence between visual fields and visual acuity.
The inability to accurately predict visual outcomes after pituitary apoplexy may be due in part to the fact that the underlying pathophysiology of the visual acuity decline is not fully understood. The purported mechanism is sudden compression of the optic apparatus directly resulting in ischemia; if this was true, though one would expect time for decompression to be an important prognosticator. Additionally, this theory does not explain the fact that patients with ischemic apoplexy also present with visual acuity deficits. 27 Furthermore, pituitary adenomas without chiasm compression have been shown to induce thinning of the retinal nerve fiber layer and this effect is potentiated for secreting adenomas, raising the possibility of an indirect contribution to visual decline via hormonal mechanism. 28
Effect of Dichotomization
To assess the effect of dichotomization would have on this population, we intentionally divided patients who underwent surgical decompression before or after several cut-off intervals. The objective of this analysis was to detect if any statistically significant differences arose after dichotomization with the knowledge that there was no actual association between timing and visual acuity recovery. Of the seven cut-off values assessed, none showed a statistically significant association with the outcome, but those patients who underwent surgery within 2 days of symptom onset had a decreased likelihood of a good visual outcome. Hence, dichotomization did not generate any spurious statistically significant associations. The trend toward worse outcomes for patients undergoing surgery within 2 days may be due to selection bias based on the surgeon's decision to treat these patients urgently, or residual bias introduced by dichotomization. As anticipated, this effect was washed out when continuous data were used to assess this relationship.
The odds ratios and correlation coefficients derived from analysis of intentionally dichotomized data did demonstrate a nonsignificant trend toward greater improvement with increasing delay from symptom onset to surgery. As evidenced by the wide CIs, this apparent relation may be simply due to the limited sample size of the study. Another potential confounder which might explain this relationship is the kinetics of visual symptom onset. Those patients with more precipitous visual decline may seek medical attention earlier (allowing for more expedient surgical management), but the rate of visual symptom onset may indicate a critical or irreparable injury to the optic apparatus.
Implications for Practice
In the context of numerous other studies indicating no benefit of early surgical intervention, the findings of this study suggest that a predefined “deadline” for surgical intervention in pituitary apoplexy may be unnecessary. That is not to say that all patients with pituitary apoplexy may be managed expectantly—on the contrary, all patients in this study underwent urgent surgery—but the results do suggest that a short delay in surgery relative to symptom onset may not put visual acuity at any substantial additional risk.
This finding is especially relevant in the situation when adherence to a predefined “deadline” would require surgery to be performed (1) before a patient is completely medically stabilized, (2) by an on-call neurosurgeon inexperienced with pituitary surgery, or (3) outside of normal working hours in settings without ideal perioperative support (i.e., the postulated “weekend effect”). 29 In these scenarios, the risk associated with having an ill-equipped operative team or suboptimal setting may theoretically outweigh the risk of waiting an additional 1 or 2 days to have the procedure done under more ideal conditions. In lieu of a fixed deadline for timing of surgery, treatment should be determined on an individual patient basis acknowledging limitations of the practice setting.
Limitations
As with any retrospective cohort study of a rare condition, this study was hampered by small sample size, absence of a control group, and susceptibility to multiple types of bias. Interpretation of these results should be tempered by the fact that the study had limited sample size and a multivariable regression could not be performed. Postoperative visual assessments were not missing completely at random, indicating potential selection bias in the assessment of outcomes. This is partly the result of our institution's large catchment area, as many out-of-state patients return to see local ophthalmology providers (or other providers within a specific insurance network). Additionally, the interval from surgery to postoperative visual assessment varied widely (data not shown) which may bias the true rate of visual recovery, as improvement may continue for years after resection of pituitary adenomas. 26
Importantly, this study cohort included only patients who underwent operative treatment for apoplexy in the acute setting. This excludes, for instance, those patients who present to clinic with a remote history of symptomatic apoplexy that have recovered vision spontaneously or remained stable for an extended period, or those who are managed conservatively for a time prior to definitive treatment of the underlying adenoma. Furthermore, there are limited rigorous data on the natural history of visual acuity in untreated pituitary apoplexy, as the standard approach for many years has been to treat this condition.
Perhaps the most critical shortcoming of this study design is probable confounding by indication, in other words, the urgency of surgical treatment was not random but determined by the surgeon's assessment which could be partially based on covariates not accounted for in this analysis. Specifically, the kinetics of visual decline such as ongoing, progressive loss of visual acuity is an important factor which is not accounted for in this analysis.
As pituitary apoplexy is a rare condition, it is not practically amenable to well-powered prospective study or randomization. Future work will necessitate aggregation of data from multiple centers to allow for larger sample size and more rigorous multivariable analysis. The results of this study suggest that dichotomous timing data may produce misleading results; for this reason, it is critical that continuously valued time delays and visual acuity be accurately measured in future collaborative work.
Conclusion
This single-center, retrospective cohort study found that there was no significant association between the continuously valued measures of time delay from symptom onset to surgical intervention and visual acuity improvement in patients suffering from pituitary apoplexy. Use of dichotomous time cutoffs or outcome assessments may suggest untrue associations.
Acknowledgments
Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under award number T32CA106183.
Funding Statement
Funding This study was funded by National Cancer Institute (Grant/Award Number: “T32CA106183”).
Footnotes
Conflict of Interest P.D.K. reports grants from National Institutes of Health—National Cancer Institute (T32CA106183), during the conduct of the study; L.B.C. reports other from Stryker, outside the submitted work.
References
- 1.Aghi M K, Bodach M E, Tumialan L M.Guidelines on the management of patients with nonfunctioning pituitary adenomas: introduction and methodologyAccesses September 17, 2019 at:https://www.cns.org/sites/default/files/guideline-chapter-pdf/nfpa-chapter-1.pdf
- 2.Briet C, Salenave S, Chanson P. Pituitary apoplexy. Endocrinol Metab Clin North Am. 2015;44(01):199–209. doi: 10.1016/j.ecl.2014.10.016. [DOI] [PubMed] [Google Scholar]
- 3.Baldeweg S E, Vanderpump M, Drake W. Society for Endocrinology Endocrine Emergency Guidance: emergency management of pituitary apoplexy in adult patients. Endocr Connect. 2016;5(05):G12–G15. doi: 10.1530/EC-16-0057. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Tu M, Lu Q, Zhu P, Zheng W. Surgical versus non-surgical treatment for pituitary apoplexy: A systematic review and meta-analysis. J Neurol Sci. 2016;370:258–262. doi: 10.1016/j.jns.2016.09.047. [DOI] [PubMed] [Google Scholar]
- 5.Abdulbaki A, Kanaan I. The impact of surgical timing on visual outcome in pituitary apoplexy: Literature review and case illustration. Surg Neurol Int. 2017;8(01):16. doi: 10.4103/2152-7806.199557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Pyrgelis E S, Mavridis I, Meliou M. Presenting symptoms of pituitary apoplexy. J Neurol Surg A Cent Eur Neurosurg. 2018;79(01):52–59. doi: 10.1055/s-0037-1599051. [DOI] [PubMed] [Google Scholar]
- 7.Forbes J A, Wilkerson J, Chambless L. Safety and cost effectiveness of early discharge following microscopic trans-sphenoidal resection of pituitary lesions. Surg Neurol Int. 2011;2:66. doi: 10.4103/2152-7806.81723. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bulters D O, Shenouda E, Evans B T, Mathad N, Lang D A. Visual recovery following optic nerve decompression for chronic compressive neuropathy. Acta Neurochir (Wien) 2009;151(04):325–334. doi: 10.1007/s00701-009-0192-x. [DOI] [PubMed] [Google Scholar]
- 9.Stata Statistical Software: Release 15 College Station, TX: StataCorp LLC; 2017. Accessed, December 15, 2019 at:https://www.stata.com [Google Scholar]
- 10.Singh T D, Valizadeh N, Meyer F B, Atkinson J L, Erickson D, Rabinstein A A. Management and outcomes of pituitary apoplexy. J Neurosurg. 2015;122(06):1450–1457. doi: 10.3171/2014.10.JNS141204. [DOI] [PubMed] [Google Scholar]
- 11.Giritharan S, Gnanalingham K, Kearney T. Pituitary apoplexy - bespoke patient management allows good clinical outcome. Clin Endocrinol (Oxf) 2016;85(03):415–422. doi: 10.1111/cen.13075. [DOI] [PubMed] [Google Scholar]
- 12.Jho D H, Biller B M, Agarwalla P K, Swearingen B. Pituitary apoplexy: large surgical series with grading system. World Neurosurg. 2014;82(05):781–790. doi: 10.1016/j.wneu.2014.06.005. [DOI] [PubMed] [Google Scholar]
- 13.Vargas G, Gonzalez B, Guinto G. Pituitary apoplexy in nonfunctioning pituitary macroadenomas: a case-control study. Endocr Pract. 2014;20(12):1274–1280. doi: 10.4158/EP14120.OR. [DOI] [PubMed] [Google Scholar]
- 14.Chen L, White W L, Spetzler R F, Xu B. A prospective study of nonfunctioning pituitary adenomas: presentation, management, and clinical outcome. J Neurooncol. 2011;102(01):129–138. doi: 10.1007/s11060-010-0302-x. [DOI] [PubMed] [Google Scholar]
- 15.Zhang X, Fei Z, Zhang W. Emergency transsphenoidal surgery for hemorrhagic pituitary adenomas. Surg Oncol. 2007;16(02):115–120. doi: 10.1016/j.suronc.2007.06.001. [DOI] [PubMed] [Google Scholar]
- 16.Bills D C, Meyer F B, Laws E R., Jr. A retrospective analysis of pituitary apoplexy. Neurosurgery. 1993;33(04):602–608. doi: 10.1227/00006123-199310000-00007. [DOI] [PubMed] [Google Scholar]
- 17.Royston P, Altman D G, Sauerbrei W. Dichotomizing continuous predictors in multiple regression: a bad idea. Stat Med. 2006;25(01):127–141. doi: 10.1002/sim.2331. [DOI] [PubMed] [Google Scholar]
- 18.Naggara O, Raymond J, Guilbert F, Roy D, Weill A, Altman D G. Analysis by categorizing or dichotomizing continuous variables is inadvisable: an example from the natural history of unruptured aneurysms. AJNR Am J Neuroradiol. 2011;32(03):437–440. doi: 10.3174/ajnr.A2425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Chen H, Cohen P, Chen S. Biased odds ratios from dichotomization of age. Stat Med. 2007;26(18):3487–3497. doi: 10.1002/sim.2737. [DOI] [PubMed] [Google Scholar]
- 20.Heavner K K, Phillips C V, Burstyn I, Hare W. Dichotomization: 2× 2 (× 2× 2× 2...) categories: infinite possibilities. BMC Med Res Methodol. 2010;10(01):59. doi: 10.1186/1471-2288-10-59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Briet C, Salenave S, Bonneville J F, Laws E R, Chanson P. Pituitary apoplexy. Endocr Rev. 2015;36(06):622–645. doi: 10.1210/er.2015-1042. [DOI] [PubMed] [Google Scholar]
- 22.Zaidi H A, Cote D J, Burke W T. Time course of symptomatic recovery after endoscopic transsphenoidal surgery for pituitary adenoma apoplexy in the modern era. World Neurosurg. 2016;96:434–439. doi: 10.1016/j.wneu.2016.09.052. [DOI] [PubMed] [Google Scholar]
- 23.Muthukumar N, Rossette D, Soundaram M, Senthilbabu S, Badrinarayanan T. Blindness following pituitary apoplexy: timing of surgery and neuro-ophthalmic outcome. J Clin Neurosci. 2008;15(08):873–879. doi: 10.1016/j.jocn.2007.08.010. [DOI] [PubMed] [Google Scholar]
- 24.Grzywotz A, Kleist B, Möller L C. Pituitary apoplexy - a single center retrospective study from the neurosurgical perspective and review of the literature. Clin Neurol Neurosurg. 2017;163:39–45. doi: 10.1016/j.clineuro.2017.10.006. [DOI] [PubMed] [Google Scholar]
- 25.Rutkowski M J, Kunwar S, Blevins L, Aghi M K. Surgical intervention for pituitary apoplexy: an analysis of functional outcomes. J Neurosurg. 2018;129(02):417–424. doi: 10.3171/2017.2.JNS1784. [DOI] [PubMed] [Google Scholar]
- 26.Gnanalingham K K, Bhattacharjee S, Pennington R, Ng J, Mendoza N. The time course of visual field recovery following transphenoidal surgery for pituitary adenomas: predictive factors for a good outcome. J Neurol Neurosurg Psychiatry. 2005;76(03):415–419. doi: 10.1136/jnnp.2004.035576. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Agrawal D, Mahapatra A K, Post K D, Ludecke D. Visual outcome of blind eyes in pituitary apoplexy after transsphenoidal surgery: a series of 14 eyes. Surg Neurol. 2005;63(01):42–46. doi: 10.1016/j.surneu.2004.03.014. [DOI] [PubMed] [Google Scholar]
- 28.Cennamo G, Auriemma R S, Cardone D. Evaluation of the retinal nerve fibre layer and ganglion cell complex thickness in pituitary macroadenomas without optic chiasmal compression. Eye (Lond) 2015;29(06):797–802. doi: 10.1038/eye.2015.35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Bell C M, Redelmeier D A. Mortality among patients admitted to hospitals on weekends as compared with weekdays. N Engl J Med. 2001;345(09):663–668. doi: 10.1056/NEJMsa003376. [DOI] [PubMed] [Google Scholar]