Outcomes After Noncardiac Surgery For Patients With Pulmonary Hypertension: A Historical Cohort Study

Abstract Background: Pulmonary hypertension (PH) is a significant preoperative risk factor. We aimed to determine predictors of perioperative morbidity and mortality after noncardiac surgery for patients with precapillary PH. Methods: We conducted a retrospective cohort study of adults with pulmonary hypertension having surgery at a single large medical referral center. The PH and surgical databases were reviewed from 2010 to 2017. Patients were excluded if PH was attributable to left-sided heart disease or they had undergone cardiac or transplant operations. To assess whether PH-specific diagnostic or cardiopulmonary testing parameters were predictive of perioperative complications, analyses were performed using generalized estimating equations. Results: Of 196 patients with PH undergoing noncardiac operations, 53 (27%) experienced 1 or more complications, including 5 deaths (3%) within 30 days. After adjustment for age and PH type, there were more complications in those undergoing moderate-to-high vs low-risk procedures (odds ratio OR, 4.17 95% CI, 2.07 to 8.40; P<0.001). After adjustment for age, surgical risk, and PH type, the risk for complications was higher for patients with worse functional status (OR, 2.39 95% CI, 1.19 to 4.78; P=0.01 for class 3/4 vs class 1/2) and elevated serum N-terminal fragment of the prohormone brain natriuretic peptide (NT-proBNP) (OR, 2.28 95% CI, 1.05 to 4.96; P=0.04 for ≥300 vs. <300 pg/mL). From an analysis that included covariates for age, surgical risk, and functional status, elevated serum NT-proBNP levels remained associated with increased risk (OR, 2.23 95% CI, 1.05 to 4.76; P=0.04). Conclusions: Patients with PH undergoing noncardiac surgery with general anesthesia have a high frequency of perioperative complications. Specific clinical (functional status), diagnostic (serum NT-proBNP), and intraoperative factors (higher-risk surgery) are predictive of worse outcomes.

Severity of PH was also assessed from clinical testing. The results of the closest test obtained prior to surgery were considered. The results of the 6-minute walk test were dichotomized (≥330 vs. <330 meters). N-terminal fragment of the prohormone brain natriuretic peptide (NT-proBNP) is a biochemical marker used in risk stratification of patients with congestive heart failure [19]. NT-proBNP was dichotomized using a cut point of <300 pg/mL which has been shown to have a 99% negative predictive value for excluding acute congestive heart failure, although this diagnostic utility is hampered by impaired renal function.
[20] Spirometry results were considered abnormal when the forced expiratory volume in the first second of expiration (FEV 1 ) was <80% of predicted, the forced vital capacity <70% of predicted, and/or diffusing capacity ≤80% of predicted.
Transthoracic 2-dimensional echocardiography (2DE) results included ejection fraction, diastolic dysfunction, right ventricular (RV) enlargement or hypertrophy, estimated RV systolic pressure (RVSP), and RVSP as a ratio of systemic systolic blood pressure (RVSP/SBP). PH was considered moderate if the RVSP/SBP was <0.66 or severe if the RVSP/SBP was ≥0.66. We also report tricuspid annular plane systolic excursion (TAPSE), a parameter of RV systolic function, which is used as a prognostic factor in PH [21,22].

Anesthesia and Surgical Characteristics
Procedures were categorized as low-risk, and intermediate to high risk operations (Table   3). We reviewed the duration of surgery, anesthetic agents used, and vasoactive pharmacologic treatments.

Outcome Measures
In addition to 30-day mortality, we considered in-hospital complications, specifically, major cardiovascular, pulmonary, neurologic, and renal complications (increased postoperative serum creatinine within 48 hours of ≥0.3 mg/dL from the preoperative value [25] or increased serum creatinine >1.3 mg/dL after surgery). We also considered all surgical complications (bleeding, newly developed coagulopathy, and reoperation for any reason); or other complications primarily related to existing pathology at the time of surgery (e.g., postoperative sepsis or septic shock in a patient undergoing surgery for gangrenous gallbladder).

Statistical Analyses
Patient and procedural characteristics were summarized for all qualifying operations using count and percentage for nominal or dichotomous variables and mean (SD) or median and interquartile range (IQR) for continuous variables. Perioperative complications were categorized as cardiovascular, pulmonary, neurologic, renal, and other. To assess whether PH-specific diagnostic or cardiopulmonary testing parameters were predictive of perioperative complications, a series of analyses were performed. Since some individuals underwent multiple surgeries during the study period these analyses were performed using generalized estimating equations (GEE). For these models, the dependent variable was any complication or death within 30 days. The GEE analyses were performed with a logit link function using robust standard error estimates. For each diagnostic and testing parameter, patients were grouped using predetermined categories. Each characteristic was assessed by univariate analysis and also after adjusting for age, surgical risk (low vs intermediate/high), and WHO group (PH 1 vs 3 vs 4). Because impaired renal function may hamper NT-proBNP diagnostic utility, we excluded patients with serum creatinine levels >2 mg/dL from the analysis of NT-proBNP. Findings from all analyses were summarized using odds ratios (OR) and corresponding 95% CI. In all cases, P<.05 was used to denote statistical significance. Analyses were performed with the R statistical package (The R Foundation) [26].

Results
During the study period, 131 individuals with PH underwent 196 operations that met study criteria. Of these, 89 individuals underwent 1 operation; 29 underwent 2 operations; 7 underwent 3 operations; 2 underwent 4 operations; and 4 underwent 5 operations.
Baseline patient data for the 196 surgeries are summarized in Table 1. The mean (SD) patient age was 59.0 (14.9) years. The comorbidity burden and use of vasoactive medications was high, as would be expected for this patient population. The diagnostic and hemodynamic variables for PH are summarized in Table 2. Of the 196 patients, 144 (74%) had group 1 PH; 27 (14%), group 3 PH; and 25 (13%), group 4 PH. The WHO functional classification was class I/II for 115 (59%) patients and class III/IV for 81 (41%).
NT-proBNP levels were available before the surgical procedure for 144 patients and were elevated (≥300 pg/mL) for 79 (55%) patients. Other echocardiographic and right-sided heart catheterization parameters were similarly reflective of a mix of patients with various stages of PH (Table 2). Procedural characteristics are described in Table 3. There were 108 (55%) low-risk and 88 (45%) intermediate or high-risk operations. Most (98.5%) operations were performed using general anesthesia with volatile agents. Pulmonary vasodilators were used during 165 (84%) cases, and vasopressor infusions were used during 161 (82%) cases (Table 3).
Fifty-three patients (27%) had 1 or more perioperative complications ( vs <300 pg/mL). No other echocardiographic or right-sided heart catheterization parameters were found to be significantly associated with complications. To assess whether elevated serum NT-proBNP level was independently predictive of complications after adjusting for functional class, an analysis was performed that included covariates for age, surgical risk, and WHO PH functional class. From this analysis, elevated NT-proBNP serum level was still found to be associated with an increased risk for complications (OR, 2.23 [95% CI, 1.05 to 4.76]; P=.04).

Emergency Procedures
A total of 13/196 (6.6%) procedures were emergent. Of these, 8 (62%) were associated with perioperative complications, and 2 (15%) patients died. In the nonemergency group, there were 183 procedures with 45 (25%) complications, including 3 (2%) that resulted in the patient's death. The difference in the rate of complications between emergent vs nonemergent procedures was significant (P=0.008). From an analysis that included only nonemergency surgery and that was adjusted for age, surgical risk, and type of PH, the risk for complications was shown to be higher for patients with worse WHO functional class (OR, 2.72 [95% CI, 1.28 to 5.78]; P=0.009 for class III/IV vs class I/II).

30-Day Mortality
Five patients (3%), 4 females and 1 male, age range between 43 and 65 years, died within 30 postoperative days ( Table 6). The limited number of deaths in our series precluded performing analyses to assess predictors of mortality.

Discussion
Among patients with different types of PH undergoing noncardiac surgery, 27% had at least 1 major complication, and 3% died within 30 days. In addition to high risk surgery, independent predictors for postoperative complications were WHO functional class III/IV, elevated serum levels of NT-proBNP.

Perioperative Morbidity and Mortality
Information regarding perioperative morbidity and mortality of patients with PH undergoing noncardiac surgery is scarce. In our earlier report [6], which used a similar study design, we reported a 42% complication rate for PH patients who underwent noncardiac surgery between 1991 and 2003. In the present study the rate of perioperative complications between 2010 and 2017 is lower, 27%. Compared with our earlier study, [6] the present cohort included a higher proportion of patients with advanced stages of PH (41% of patients were in WHO functional class III/IV in the present study vs 27% of patients in the earlier study[6] who were in New York Heart Association functional class III and IV). At the same time, in the present study, 45% of patients underwent high-risk procedures compared with 79% in the earlier study [6]. PH functional class and exposure to higher-risk procedures are the 2 main determinates of surgical outcomes in patients with PH, and because these risks were divergently distributed between the 2 study epochs, it is difficult to comment as to whether the decreased rate of complications in the present study was related to advances in management over the decades or to an imbalance of risk factors between the 2 cohorts.

Predictors of Morbidity
Price et al. [28] reported that risk factors for perioperative complications in patients with PH undergoing noncardiac surgery were greater for emergency, major, and longer surgery.
In the present study, we confirmed that the acuity of surgery is an important predictor of morbidity, and patients undergoing intermediate/major operations had 4.2 times increased likelihood for experiencing a perioperative complication than patients undergoing minor procedures. Furthermore, patients in functional class III/IV had a 2.4 times higher likelihood of experiencing a complication than patients in class I/II. This mirrors our earlier finding that New York Heart Association class ≥II in patients with PH was associated with an odds ratio of 2.9 for development of postoperative complications [6]. Finally, in the present study, only 13 patients underwent an emergency operation, and they had higher complication (62%) and mortality (15%) rates than patients undergoing nonemergency procedures.
NT-proBNP is released from ventricular myocytes in response to mechanical stretching related to increased ventricular pressure or volume overload and has been used as a surrogate marker for ventricular failure [29]. In our study, an elevated NT-proBNP level was a predictor of perioperative complications. Specifically, an NT-proBNP serum level ≥300 ng/mL was associated with 2.2-fold higher odds for developing a complication even after we adjusted for age, type of surgery, and WHO functional class.
Different functional tests, echocardiographic variables, and cardiac catheterization parameters were formerly used to predict long-term outcomes for nonsurgical patients with PH [22,30,31]. For example,Miyamoto et al. [31] demonstrated that the 6-minute walk test was predictive of mortality for patients with primary PH during a mean follow-up period of 21 months. Forfia et al. [22] showed TAPSE to be a good long-term predictor of mortality for patients with PH. Several studies showed that abnormal 6-minute walk and 2DE parameters were predictive of increased perioperative morbidity and mortality [11,32]. For example, TAPSE <18 mm was found to be an independent predictor of intraoperative cardiopulmonary resuscitation and death in patients undergoing emergent pulmonary embolectomy [21]. In our study, none of the selected parameters derived from the pulmonary and cardiac tests was found to be a statistically significant predictor of perioperative morbidity, which is similar to findings from our previous study [6]. However, it is noteworthy that for pulmonary function tests, echocardiography, and right-sided heart catheterization variables (RV enlargement, RVSP/SBP ratio, RV strain, TAPSE), the point estimates are directionally consistent with the premise that worse PH stage may be associated with a higher likelihood for perioperative complications. Since these measurements were not available for all patients the statistical power for these analyses was limited. Therefore, the results for these parameters should not be interpreted as evidence of no association.

Limitations
Our study has inherent limitations related to its retrospective design. Both known and unknown confounding variables may have been present and thus not accounted for in the analyses. We have identified variables associated with morbidity; however, we cannot conclude causality. In this study we included a wide variety of surgical procedures, which implies vastly different surgical acuities. Due to the limited overall sample-size, we cannot perform analyses which adjust for specific surgery types.

Conclusions
In conclusion, patients with PH undergoing noncardiac surgery under general anesthesia had a high frequency of perioperative complications. The main predictors of perioperative morbidity included poor baseline functional status, increased serum NT-proBNP, and higher-risk operations.