Course, Predictors of Diaphragm Recovery After Phrenic Nerve Injury During Pediatric Cardiac Surgery

Transcription

1 Course, Predictors of Diaphragm Recovery After Phrenic Nerve Injury During Pediatric Cardiac Surgery Brendan M. Smith, MD, Nchedochukwu J. Ezeokoli, Alaina K. Kipps, MD, Anthony Azakie, MD, and Jeffery J. Meadows, MD Pediatric Division of Cardiology and Department of Surgery, University of California, San Francisco, San Francisco, California Background. Hemidiaphragm paralysis from phrenic nerve injury is a known complication of congenital cardiac surgery. Return of diaphragm function has been reported; however, prior studies on this subject have been limited by small numbers, static assessment methods, or observation of plicated or non-plicated patients alone. To describe return of function, we reviewed fluoroscopy and ultrasonography in all diagnosed cases of diaphragmatic paralysis. Methods. Surgical cases at our institution between 1991 and 2010 were identified for patients with postoperative hemidiaphragm paralysis diagnosed by chest X-ray, ultrasound, or fluoroscopy. Follow-up ultrasound and fluoroscopic studies were reviewed for return of diaphragm function. Results. Seventy-two cases of postoperative hemidiaphragm paralysis were identified. Forty cases were plicated prior to discharge. Plicated patients were younger at time of diagnosis (median 46 days average 3.6 months; p [ 0.025) and had a larger proportion of single ventricle diagnoses (48% vs 16%) compared with non-plicated patients. Twenty-six patients with paralysis were excluded in follow-up due to lack of studies documenting diaphragm function after the diagnostic study. Of the remaining 46 cases, median follow-up was 353 days (range: 6 days to 17 years). Plicated and nonplicated patients regained function at similar frequency (60% and 54.8%, respectively). Plication status, Risk Adjustment for Congenital Heart Surgery (RACHS) 1 score, age at diagnosis, and side of paralysis did not predict failure of recovery. Conclusions. In the current era, return of diaphragm function after phrenic nerve injury sustained during congenital cardiac surgery is a known occurrence; however, predicting recovery continues to be difficult. (Ann Thorac Surg 2013;96:938 42) Ó 2013 by The Society of Thoracic Surgeons Phrenic nerve injury resulting in transient or permanent diaphragmatic dysfunction is a well-known complication of surgical intervention for congenital heart disease. Symptomatic diaphragmatic dysfunction typically presents in the postoperative period with failure to wean from ventilatory support, or hypercarbia, persistent atelectasis, and respiratory distress after extubation. Neonates and infants are especially vulnerable to the effects of diaphragmatic dysfunction due to supine positioning, dependence upon the diaphragms for optimal respiratory effort, and the increased relative mobility of mediastinal and thoracic anatomy in infancy [1 5]. Several studies have demonstrated the early benefits of plication with respect to facilitating extubation and shortening intensive care unit stay in this patient population [3, 6 9]. However, the early benefits of plication must be weighed against the knowledge that spontaneous recovery of diaphragm function can occur. Unfortunately, our knowledge about the phenomena of diaphragmatic recovery remains rudimentary. It is clear that recovery Accepted for publication May 17, Address correspondence to Dr Smith, 505 Parnassus Ave, M696, Box 0110 San Francisco, CA 94143; smithb@peds.ucsf.edu. can occur months or years after the initial insult [10 14], but the time course and predictors of recovery are unknown. We sought to clarify the course of spontaneous recovery in patients sustaining diaphragmatic dysfunction after surgery for congenital heart disease. Patients and Methods Patient Data Records of congenital heart surgery cases performed at the University of California, San Francisco between the years of 1991 and 2010 were reviewed for patients diagnosed with diaphragm paralysis by chest X-ray, fluoroscopy, or postoperative echocardiogram. Patients who had suffered paralysis due to procedures done at other institutions prior to care at the University of California of San Francisco and those with incomplete postoperative records were excluded. Cases of bilateral paralysis were also excluded, as this was felt to possibly represent a separate clinical entity with a different prognosis for recovery. Charts were reviewed for demographic data, diagnosis, cardiac anatomy, type of surgery performed, side of diaphragm paralysis, and postoperative course with regard to time to diagnosis and plication. The Committee for Ó 2013 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc

2 Ann Thorac Surg SMITH ET AL 2013;96: DIAPHRAGM RECOVERY AFTER CARDIAC SURGERY 939 Human Rights at the University of California of San Francisco approved the study. Return of Diaphragm Function Data regarding return to function wee gathered by nonblinded review of all available postoperative and followup fluoroscopic and echocardiographic studies that interrogated diaphragm movement during spontaneous respiration. Patients with studies demonstrating elevated, immobile, or paradoxically moving diaphragms were considered to have failed return of diaphragm function. Patients without adequate diaphragmatic imaging after the initial diagnosing paralysis were censored from the analysis at the time of the last dynamic imaging study. Statistical Analysis Descriptive data were expressed as mean and standard deviation or median and range as appropriate to their distributions. Comparisons between unpaired continuous variables with an appropriate distribution were made using the unpaired t test. Estimations of time-dependent outcomes were made using Kaplan-Meier analysis, and analyses of time-dependent outcomes between groups were made utilizing the log-rank test. Risk factor analysis of dichotomous outcomes was made using logistic regression. Results Patient Characteristics Ninety patients with diaphragmatic dysfunction were identified. Sixteen patients had inadequate documentation to be included in the study. One patient died shortly after surgery from sepsis and 1 patient was transferred postoperatively for heart transplantation. The remaining 72 patients had available perioperative data and form this study group. The most common surgical procedures in these patients were the following: those who included placement or take-down of a Blalock-Taussig shunt (27.8%); bidirectional cavopulmonary anastomosis (19.4%); and the arterial switch procedure (15.3%). Thirtynine percent of cases occurred during reoperations, the remaining 61% occurred during an initial surgical intervention (Table 1). A postoperative echocardiogram was the initial method of diagnosis in 51 patients, while fluoroscopy and chest X-ray accounted for 14 and 7 diagnoses, respectively. The median time to diagnosis for all cases was Table 1. Procedure Leading to Diaphragm Paralysis Procedure Number of Procedure Percentage of DP Cases Reoperations Within Procedure Group Bidirectional Glenn Arterial switch VSD closure, RV to PA conduit Norwood with BT shunt Norwood with Sono modification VSD closure, RVOT resection Aortic arch reconstruction ASD/VSD closure Coarctation repair Blaylock-Taussig shunt VSD closure, RV to PA conduit, unifocalization VSD closure, RVOT patch augmentation TAPVR repair Truncus arteriosus repair with RV to PA conduit Unifocalization, unilateral Takedown of bidirectional Glenn Arterial and atrial switch Hybrid Norwood VSD closure Ross-Konno Fontan take-down Fontan PA banding Total cases of diaphragm paralysis ASD ¼ atrial septal defect; BT ¼ Blalock-Taussig; DP ¼ diaphragm paralysis; PA ¼ pulmonary artery; RV ¼ right ventricle; RVOT ¼ right ventricular outflow track; TAPVR ¼ total anomalous pulmonary venous return; VSD ¼ ventricular septal defect.

3 940 SMITH ET AL Ann Thorac Surg DIAPHRAGM RECOVERY AFTER CARDIAC SURGERY 2013;96: days (0 to 368 days). Unilateral dysfunction occurred on the left side in 40 cases, while 30 cases involved the right diaphragm. Perioperative Period and Diaphragm Plication Forty of 72 (56%) patients underwent diaphragm plication prior to discharge. The indications for plication were failure to wean from mechanical ventilation or significant respiratory compromise after extubation. The median time from diagnosis to plication was 3 days (range 0 to 37 days). Patients who were plicated tended to be younger with a median age at time of diagnosis of 46 days (8 days to 4.2 years) compared with a median age 3.6 months (7 days to 10.7 years) for non-plicated patients (p ¼ 0.025). Patients undergoing plication were also more likely to have single ventricle physiology (48% vs 16%), the most common of which was hypoplastic left heart syndrome (10 of 19 patients or 53%). Follow-Up Period Clinical follow-up and vital status were available for 69 patients, of whom 46 had adequate dynamic imaging to assess diaphragm function. Median follow-up was 353 days (6 days to 17 years) in the 69 patients with available clinical documentation. During the follow-up period, 7 patients died, all of whom had single ventricle anatomy and were interstage post-norwood mortalities. Three of these 7 had the diagnosis of hypoplastic left heart syndrome, and 5 had severe valvular insufficiency. The Risk Adjustment for Congenital Heart Surgery (RACHS) score greater than 5 (hazard ratio 3.8) and diaphragm plication (hazard ratio 4.2) were associated with increased risk of mortality but did not reach statistical significance (p ¼ 0.12 and p ¼ 0.12, respectively). Of the 46 patients with adequate dynamic imaging, to assess late diaphragm function follow-up was 466 days (2 days to 13 years), and late imaging was more commonly available in plicated patients (31 of 38, 82%) than non-plicated patients (15 of 31, 48%), p ¼ Diaphragmatic recovery occurred in 26 (56.5%) patients and was similar between plicated and non-plicated patients; 60% and 54.8%, respectively. All 46 patients (including those who did not recover function within the first year during follow-up) had at least 1 study within the first 12 months after discharge. Median time to the first study performed after discharge was 53 days (range 4 to 189 days) and the average number of studies within the first 12 months after diagnosis was 4 (range 1 to 29). All patients eventually had more than 1 study after the initial diagnosis and those with resolution had at least 1 prior study showing persistent dysfunction after initial diagnosis. The median number of studies each patient received during duration of clinical follow-up reported in this series was 5 and the number of studies ranged between 1 and 33. This occurred over a median follow-up period of 353 days, with a range between 6 days to 17 years. Fluoroscopy was the study diagnostic of return of function in 15 out of the final 26 instances. Kaplan- Meier estimates of diaphragm recovery demonstrated a steady rate of recovery for several years from surgery (Fig 1). Analysis of risk factors for failure of diaphragmatic recovery showed that neither age at surgery, side of diaphragm dysfunction, RACHS score, or plication status predicted failure (Table 2). Comment Phrenic nerve injury resulting in diaphragmatic dysfunction is a well-known complication of surgical intervention for congenital heart disease. While the prevalence and predictors of diaphragm dysfunction in this setting have been well characterized, this study is the Fig 1. Recovery of diaphragm function over follow-up period. Recovery occurred over an interval between the prior study to demonstrate dysfunction and the last study demonstrating resolution of paresis. The curve in red estimates recovery as an event occurring at the beginning of the interval (time of prior study showing dysfunction), the blue curve estimates recovery as an event occurring at the time of the study demonstrating function, and the green as a mid-point estimate between the 2. (x axis ¼ time in months after surgery; y axis ¼ proportion of patients with ongoing diaphragm dysfunction.)

4 Ann Thorac Surg SMITH ET AL 2013;96: DIAPHRAGM RECOVERY AFTER CARDIAC SURGERY 941 Table 2. Analysis of Predictors of Recovery Predictor Odds Ratio p Value 95% CI Age at surgery Diaphragm plication Diaphragm side RACHS 1 score CI ¼ confidence interval; Heart Surgery. RACHS ¼ Risk Adjustment for Congenital first to our knowledge to analyze both the time course and predictors of diaphragm recovery. The majority of patients (56.5%) in our series demonstrated return of diaphragm function, with a steady recovery rate over the first several years. Unfortunately, neither age at surgery, affected diaphragm side, plication status, nor RACHs score predicted recovery. Prior studies on this subject have been limited by small numbers, static assessment methods, or limited observation to a selected subset of patients. Lemmer and colleagues [15] assessed diaphragm function in the mid-term period in postoperative plicated and nonplicated patients using ultrasound. Out of 23 patients, a minority (2 of 11 plicated diaphragms and 6 of 16 non-plicated diaphragms) demonstrated a return of function; however, these results are difficult to interpret as 4 patients with bilateral paralysis were counted as 2 cases of unilateral paralysis each. We chose to exclude 2 patients with bilateral paralysis as it was felt this may represent a separate clinical entity. Of note, both cases of bilateral paralysis resolved in less than 60 days. Baker and colleagues [10] followed 17 patients with unilateral paralysis post-plication at variable intervals with fluoroscopy. The majority (16 of 17) of their patients demonstrated return of function; however, their analysis was limited to plicated patients exclusively. The remainder of the studies have relied on either clinical symptoms or static chest X- rays as indicators of recovery [3 6, 8, 12]. The remainder of our findings are consistent with prior publications. Characteristics of this patient cohort with respect to anatomic diagnosis and early clinical outcomes were similar to those previously reported. The surgical procedure most commonly associated with diaphragm paralysis was the creation of the bidirectional Glenn, accounting for 19.4% of total cases. This is similar to the findings of Baker and colleagues [10] who noted that in their experience creation of a bidirectional Glenn was the procedure associated with the highest incidence of diaphragm dysfunction for patients undergoing single ventricle palliations. An associated procedural component of this operation is the removal of a Blalock- Taussig shunt. In our review, 27.8% of cases of diaphragm dysfunction resulted after surgeries that involved the creation or removal of a Blalock-Taussig shunt, a procedure that has been previously associated with phrenic nerve injury [4 6]. Finally, patients in our series who had undergone previous surgery prior to the procedure that resulted in their diaphragm paralysis represented 39% of our cases of diaphragm paralysis. Joho-Arreloa and colleagues [6], along with de Leuww and colleagues [5], found a similar proportion; 35% and 49%, respectively. Consideration of these findings should include acknowledgement of the limitations of our study design. First, during the period covered by this study patients underwent a chest X-ray and echocardiogram prior to discharge; however, it was common for the echocardiogram at discharge to lack subcostal views adequate to ascertain diaphragm function. As such, we may have failed to detect asymptomatic diaphragm dysfunction. More importantly, because follow-up assessment of diaphragm recovery was not standardized, our characterization of time-dependent recovery should be considered a relatively rough estimate. As such, while the data are convincing for a steady rate of return of function, a prospective program would be required for greater precision. Finally, the focus of this study was limited to describing the time course and predictors of diaphragm recovery, and while this study provides data useful for clinical decision making, data informing specific decisions on the need for and timing of diaphragm plication in the setting of postoperative diaphragmatic dysfunction remain needed. In conclusion, a significant proportion of patients who sustain unilateral diaphragm paralysis while undergoing procedures to palliate congenital heart disease will eventually demonstrate recovery of diaphragm function in the postoperative period. There appears to be a steady rate of recovery over the first several years, but predicting which patients will experience return of diaphragmatic function remains difficult. Documentation of diaphragm function in postoperative patients should be included in the routine echocardiography of these patients, at least until return of function is seen, which may occur up to several years after the initial injury. References 1. Commare MC, Kurstjens SP, Barois A. Diaphragmatic paralysis in children: a review of 11 cases. Pediatr Pulmonol 1994;18: Muller NL, Bryan AC. Chest wall mechanics and respiratory muscles in infants. Pediatr Clin North Am 1979;26: van Onna IE, Metz R, Jekel L, Sr Woolley, van de Wal HJ. Post cardiac surgery phrenic nerve palsy: value of plication and potential for recovery. Eur J Cardiothorac Surg 1998;14: T onz M, von Segesser LK, Mihaljevic T, Arbenz U, Stauffer UG, Turina MI. Clinical implications of phrenic nerve injury after pediatric cardiac surgery. J Pediatr Surg 1996;31: de Leeuw M, Williams JM, Freedom RM, Williams WG, Shemie SD, McCrindle BW. Impact of diaphragmatic paralysis after cardiothoracic surgery in children. J Thorac Cardiovasc Surg 1999;118: Joho-Arreola AL, Bauersfeld U, Stauffer UG, Baenziger O, Bernet V. Incidence and treatment of diaphragmatic paralysis after cardiac surgery in children. Eur J Cardiothorac Surg 2005;27: Schwartz MZ, Filler RM. Plication of the diaphragm for symptomatic phrenic nerve paralysis. J Pediatr Surg 1978;13:

5 942 SMITH ET AL Ann Thorac Surg DIAPHRAGM RECOVERY AFTER CARDIAC SURGERY 2013;96: Tsugawa C, Kimura K, Hishijima E, Muraji T, Yamaguchi M. Diaphragmatic eventration in infants and children: is conservative treatment justified? J Pediatr Surg 1997;32: Hamilton JRL, Tocewicz K, Elliott MJ, de Leval M, Stark J. Paralysed diaphragm after cardiac surgery in children: value of plication. Eur J Cardiothorac Surg 1990;4: Baker CJ, Boulom V, Reemtsen BL, Rollins RC, Starnes VA, Wells WJ. Hemidiaphragm plication after repair of congenital heart defects in children: quantitative return of diaphragm function over time. J Thorac Cardiovasc Surg 2008;135: Kizilcan F, Tanyel FC, Hiçs onmez A, B uy ukpamukçu N. The long-term results of diaphragmatic plication. J Pediatr Surg 1993;28: Watanabe T, Trusler GA, Williams WG, Edmonds JF, Coles JG, Hosokawa Y. Phrenic nerve paralysis after pediatric cardiac surgery. J Thorac Cardiovasc Surg 1987;94: Stone KS, Brown JW, Canal DF, King H. Long-term fate of the diaphragm surgically plicated during infancy and early childhood. Ann Thorac Surg 1987;44: Mickell JJ, Oh KS, Siewers RD, Galvis AG, Fricker FJ, Mathews RA. Clinical implications of postoperative unilateral phrenic nerve paralysis. J Thorac Cardiovasc Surg 1978;76: Lemmer J, Stiller B, Heise G, et al. Mid-term follow-up of patients with diaphragmatic plication after surgery for congenital heart disease. Intensive Care Med 2007;33: INVITED COMMENTARY Phrenic nerve injury during repair of congenital heart defects can cause serious morbidity, particularly in neonates and infants, who will usually require diaphragm plication. This report by Smith and associates [1] has added considerably to our understanding of this complication and the subsequent natural history of diaphragm function in patients with and without plication. The study is less than optimal because of the lack of a protocol for postdischarge monitoring. However, this is the best information to date because of the large number of patients, the fact that those with resolution of paresis had at least one outpatient study showing persistence of diaphragm dysfunction, the frequency of imaging in the follow-up period, and the duration of follow-up. By showing that there was equal return of function for patients with and patients without plication, this study also helps answer the concern that plication sutures in the diaphragm might themselves lead to further phrenic injury and a reduced chance of functional recovery. As in other reports on phrenic injury, bidirectional Glenn emerges as the most frequently associated procedure. That is likely related to the need for mobilization of the superior cava in the setting of reoperative adhesions rather than being associated with takedown of a previously placed shunt as suggested in the manuscript. The next most commonly associated procedure was arterial switch, in which extensive mobilization of the branch pulmonary arteries to facilitate the Le Compte maneuver likely adds to the risk. We also believe that thermal injury to either phrenic nerve from electrocautery used during removal of the thymus, a routine practice to improve exposure in neonates and infants undergoing complex repairs, is a common cause for hemidiaphragm paresis. Although our understanding of the natural history of recovery after phrenic injury has been improved by this report, it is far from complete. A prospective protocol for follow-up of patients with phrenic injury that included standardized imaging would result in much better data. We also might be able to quantitate the return of diaphragm function, an issue that was not addressed in this study. Pulmonary function testing where feasible or possibly measuring volumetric changes within the pleural space during the phases of respiration by magnetic resonance imaging or computed tomography might help address this question. Winfield Wells, MD Division of Cardiothoracic Surgery The Heart Institute at Childrens Hospital, LA Mail Stop Sunset Blvd Los Angeles, CA wwells@chla.usc.edu Reference 1. Smith BM, Ezeokoli NJ, Kipps AK, Azakie A, Meadows JJ. Course, predictors of diaphragm recovery after phrenic nerve injury during pediatric cardiac surgery. Ann Thorac Surg 2013;96: Ó 2013 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc