Bacteremia, Endocarditis, and the Hancock Valve

Transcription

1 Bacteremia, Endocarditis, and the Hancock Valve Donald J. Magilligan, Jr., M.D., Edward L. Quinn, M.D., and Julio C. Davila, M.D. ABSTRACT Among 373 patients with porcine xenografts, there were 27 instances of exposure of the xenograft to bloodstream or endocardia1 infection in 22 patients. Nine patients underwent 10 separate insertions of xenografts for active infective endocarditis. There were no early infections or valve failures. Three patients returned with a late prosthetic valve endocarditis (PVE) due to a new infection. There were 6 instances of bacteremia early after xenograft valve insertion with no early infection, no valve dysfunction, and 1 instance of late PVE. Eleven patients had PVE on a porcine xenograft. Blood cultures in the 10 patients treated with antibiotics promptly became negative. There were 3 valve-related deaths: 2 from valve incompetence and 1 from mitral and aortic xenograft stenosis. Our experience suggests that the Hancock porcine xenograft is: (1) as resistant to infection as are rigid prostheses in active infective endocarditis; (2) resistant to early postoperative bacteremias; and (3) easier to sterilize than rigid prostheses and more durable than other tissue valves in the face of PVE. The Hancock porcine xenograft valve has demonstrated satisfactory hemodynamic characteristics, a low thromboembolism rate, and excellent durability in the six years it has been in clinical use. In series reporting clinical experience with the Hancock valve there have been 7 instances of prosthetic valve endocarditis, 4 of which were successfully treated medically [5, 7, 17,281. To further examine the durability of the Hancock valve in the face of infection, we reviewed the Henry Ford Hospital experience with 470 Hancock valves inserted in 373 patients. From the Division of Thoracic and Cardiac Surgery and Department of Infectious Diseases, Henry Ford Hospital, Detroit, MI. Presented at the Thirteenth Annual Meeting of The Society of Thoracic Surgeons, Jan 24-26, 1977, San Francisco, CA. Address reprint requests to Dr. Magilligan, Division of Thoracic and Cardiac Surgery, Henry Ford Hospital, 2799 W Grand Blvd, Detroit, MI Materials and Methods Of the 373 patients with Hancock porcine xenograft valves, there were 22 in whom the xenograft valve was exposed to bloodstream or endocardial infection. Three patients had 2 episodes of infection and 1 patient had 3 infectious episodes, which account for 27 instances of exposure of the xenograft to active infection. Patients were exposed to three types of infection. The first type of exposure was insertion of a porcine xenograft in active infective endocarditis. Endocarditis was considered active if the valve culture was positive at the time of operation or if the patient had received less than four weeks of antibiotic therapy. The second type of infection was bacteremia early after insertion of a xenograft valve. These patients had fever and positive blood cultures in the first two postoperative weeks. It is uncertain whether these early septic episodes actually represent prosthetic valve infection [201; thus they are called early bacteremias. The third type of infection was prosthetic valve endocarditis (PVE) on a porcine xenograft. Diagnosis was made by 2 positive blood cultures (9 patients), positive culture of a mycotic embolus (1 patient), or at postmortem examination (1 patient). All elective patients received a cephalosporin antibiotic starting the evening before operation and continuing for 7 days. Patients with known primary or secondary endocarditis had antibiotics and dosages chosen according to sensitivities by standardized disc testing and minimum inhibitory concentrations determined by a double agar plate or tube dilution method. From October, 1971, to October, 1974, valves were rinsed in a saline solution to which neomycin and bacitracin had been added. Since that time valves have been rinsed in Ringer s lactate solution alone. Antibiotic soaking was not a factor in the xenograft s resistance to infection. 508

2 509 Magilligan, Quinn, and Davila: Bacteremia, Endocarditis, and the Hancock Valve Results Porcine Xenograft lnsertion in Active lnfective Endocarditis Nine patients underwent valve replacement with a porcine xenograft during active infective endocarditis, and one patient had valve replacement for active infection on 2 occasions, accounting for 10 separate operative procedures in these 9 patients. There were 8 men and 1 woman, whose ages ranged from 25 to 55 years. Four of the 9 were drug addicts. Valve cultures were positive in 8 of the 10 operations (organisms are listed in Table 1). The 2 patients with negative valve cultures underwent operation because of hemodynamic deterioration after 20 and 23 days of antibiotic treatment, respectively. Isolated aortic valve replacement was performed on 5 occasions, aortic and mitral valve replacement on 3 occasions, mitral and tricuspid replacement once, and isolated mitral valve replacement once. There was 1 operative death, and follow-up was possible in 8 patients who had 12 porcine xenograft valves inserted in the face of active infection. In no patient did the valve become infected during the current episode, nor was there any valve dysfunction. Five of the 8 patients who survived operation have normally functioning valves and are free of infection from 11 to 48 months postoperatively. Three patients had late infection of a porcine xenograft. Patient 1 returned with prosthetic valve endocarditis due to Streptococcus faecalis 32 months following mitral and tricuspid valve replacement. The initial infecting organism had been Candida paropsilosis. Patient 4 returned with aortic I'VE 12 months postoperatively. Although the organism (Streptococcus ) was the same as in the original infection, it was considered a new infection because of the interval between episodes of endocarditis. Patient 7 underwent first aortic and then mitral valve replacement for 2 separate episodes of active infective endocarditis. Three months after the aortic Table 1. Porcine Xenograft Valve lnsertion in Active lnfective Endocarditis Antibiotic Valve Opera- Postoperative Duration Patient Source Organism Culture tion Antibiotic (days) Follow-up 1 Addict 2 Septic arthritis 3 Unknown 4 Unknown 5 Addict 6 Addict 7 Addict Addict 8 Unknown 9 Unknown Candida + paropsilosis Propionibacterium + acne Corynebacterium; + Propionibacterium acne Streptococcus + Candida + paropsilosis; Co ynebacterium staphylococcus - aureus Streptococcus + Streptococcus + ; Propionibacterium acne Staphylococcus + epidermidis Streptococcus - MVR, TVR AVR An, MVR AVR AVR AVR, MVR AVR MVR AVR An, MVR Amphotericin B Clindamycin Cephalothin Vancomycin Amphotericin B, 5-fluocytosine, vancomycin Methicillin Penicillin G Ampicillin Methicillin None mo 48 mo 35 mo PVE at 12 mo 20 mo 11 mo PVE at 3 mo Sudden death, 4 mo 23 mo Operative death MVR = mitral valve replacement; TVR = tricuspid valve replacement; AVR = aortic valve replacement; PVE = prosthetic valve endocarditis.

3 510 The Annals of Thoracic Surgery Vol 24 No 6 December 1977 valve replacement, and 8 months prior to his mitral valve endocarditis, he had an episode of polymicrobial prosthetic valve endocarditis, which was successfully treated medically. This infection and that involving the mitral valve were considered new infections because of a long symptom-free period, new organisms, and the appearance of septicemia directly related to the reinstitution of drug abuse. This patient had no evidence of valve dysfunction or infection when he died suddenly at home, 4 months after mitral valve replacement. His family assured us he was asymptomatic 2 hours prior to his death; however, postmortem examination was not performed. These 3 patients (Nos. 1, 4, and 7) are included in the experience with prosthetic valve endocarditis occurring on a Hancock valve (see Table 3). Bacteremia Early After lnsertion of Porcine Xenograft Six patients had fever and positive blood cultures from 2 to 13 days following valve replacement with a porcine xenograft. Five of the 6 patients were women whose ages ranged from 41 to 60 years. Five patients underwent operation for rheumatic valvular disease and 1 for mitral incompetence due to a prolapsing valve. No cardiac or extracardiac signs of endocarditis were noted in any patient. Three of 6 patients were treated with specific antibiotics for six weeks, and 3 for no more than two weeks. Patient ll (Table 2) returned 21 months postoperatively with PVE from the same organism and was successfully treated medically. He is included in the experience with PVE on a Hancock porcine xenograft (Table 3). Five patients have been followed from 21 to 38 months with no evidence of residual infection or valve dysfunction. PVE on a Porcine Xenograft Eleven patients, 6 men and 5 women aged 22 to 72 years, sustained prosthetic valve endocarditis following insertion of a porcine xenograft. The interval between valve operation and endocarditis ranged from 3 to 45 months. Patients 1, 4, and 7 (see Table 3) had undergone valve replacement for active infective endocarditis; the PVE was considered a new infection. Patient 11 had an early bacteremia due to Enterobacter cloacae and returned with PVE due to the same organism 21 months later. The long interval between bacteremia and PVE suggests a new infection. Infecting organisms are listed in Table 3. The source of PVE was drug abuse in 4 patients, dental extraction in 1, urinary tract infection in 1, and unknown in 5. Patient 22 had candida on the porcine xenograft. This was discovered unexpectedly on Table 2. Bacteremia Early After Porcine Xenograft Valve lnsertion Postop Interval Duration Patient Operation Organism (days) Antibiotic (days) FoIIow-u~ 10 MVR Strep tococcus 5 Ampicillin, 35 faecalis streptomycin 38 mo 11 MVR Enterobacter 2 Gentamicin, 42 PVE, 21 mo cloacae cephalothin 12 AVR, Streptococcus 10 Ampicillin 35, 42 MVR faecalis 26 mo 13 MVR Staphylococcus 5 Vancomy cin, 14, 3 aureus clindamycin 21 mo 14 MVR Serratia 13 Gentamicin 8 marcescens 22 mo 15 MVR, Bacteroides 10 None... TVR fragilis 23 mo MVR = mitral valve replacement; PVE = prosthetic valve endocarditis; AVR = aortic valve replacement; TVR = tricuspid valve replacement.

4 511 Magilligan, Quinn, and Davila: Bacteremia, Endocarditis, and the Hancock Valve Table 3. Prosthetic Valve Endocarditis in a Porcine Xenograft Postop Opera- Interval Antibiotic Patient tion Organism Source (no) Antibiotic Duration Outcome Follow-up 1 MVR, TVR 4a AVR 7= AVR llb MVR 16 MVR 17 MVR 18 AVR 19 MVR 20 MVR 21 AVR, MVR 22 MVR Strep tococcus faecafis Strep tococcus Strep tococcus, Staphylococcus epidermidis, Hemophilus influenzae Enterobacter cloacae Streptococcus faecalis Streptococcus Streptococcus Streptococcus Strep tococcus faecafis Streptococcus faecalis Candida Addict Unknown Addict Unknown Unknown Addict Dental extraction Addict Unknown Unknown Urinary tract Ampicillin, 6 wk Penicillin, 10 days sheptomydn Ampicillin, 6wk methicillin Carbenicillin, 6 wk Penicillin, 6 wk streptomycin Penicillin, 6 wk Streptomycin Penicillin, 4 wk Ampicillin, 6 days Ampicillin, 6 wk Ampicillin 6 wk None Operation Operation Died mo Died 19 days postop MVR, 8 mo; died suddenly, 11 mo 28 mo Died, 20 mo; valve normal at postmortem examination 31 mo Operative death Drug overdose; valve normal at postmortem examination 3 mo Died, 3 mo; aortic and mitral xenograft stenosis Died, pneumonia; candida on valve at postmortem examination "Original xenograft insertion for active infective endocarditis (see Table 1) bbacteremia early postoperatively (see Table 2). MVR = mitral valve replacement; TVR = tricuspid valve replacement; AVR = aortic valve replacement. postmortem examination. Thus 10 of 11 patients who had specific antibiotic therapy for PVE had blood cultures that promptly became negative. There were 3 valve-related deaths in the 11 patients. Patient 4 underwent emergency operation for overwhelming aortic valve incompetence after 10 days of antibiotic therapy for Streptococcus PVE. At operation there was a sterile annular abscess with nearly total dehiscence of the xenograft. There were sterile vegetations on the prosthetic leaflets, but they did not appear to affect the leaflet function (Fig 1). The patient died suddenly 19 days postoperatively with an uninfected, normally functioning porcine xenograft. The second valve-related death was that of Patient 18, who was operated on for aortic incompetence due to Streptococcus PVE. At operation there was a sterile annular and subannular abscess with destruction of leaflet tissue and vegetations. The patient did not survive aortic valve replacement and Dacron patch grafting of the annular and subannular aneurysm. The third valve-related death was that of a 72-year-old woman (Patient 21) who remained in persistent congestive heart failure following successful medical treatment of Streptococcus faecalis endocarditis that involved her aortic and mitral xenograft valves. Cardiac catheterization revealed a peak systolic gradient of 30 mm Hg across the 23 mm xenograft aortic valve and a mean diastolic gradient of 30 mm Hg across the 31 mm xenograft mitral valve. The patient died of congestive heart failure 3 months after successful treatment of PVE. On postmortem examination the orifices of the aortic and mitral xenografts were partially occluded by sterile vegetations (Fig 2). Among the remaining 8 patients with PVE, no deaths were directly related to valve dysfunction, persistent infection, or embolus. Four are well after medical treatment alone for 3 to 31 months. Four others are dead; of these, 1 patient (No. 22) had unsuspected candida PVE discovered on postmortem examination (Fig 3). Patient 19 died suddenly after 8 days of antibiotic therapy for mitral xenograft PVE. The cause of death was most likely a self-inflicted drug overdose. There was no evidence of active infection or prosthetic leaflet destruction on postmortem examination. Patient 16 died 20 months after

5 512 The Annals of Thoracic Surgery Vol 24 No 6 December 1977 Fig 1. Aortic valve of Patient 4 with nearly total dehiscence of the valve, which is infected with Streptococcus. On the left it can be seen that the sutures have torn loose fragments of the annular tissue. There are vegetations on the ventricular (left) and aortic (right) sides of the leaflets, but the leaflets are intact and open and coapt normally. Fig 2. Prosthetic aortic (left) and mitral (right) valves from Patient 21, who, after successful medical treatment of prosthetic valve endocarditis, had30 mm Hggradients across the valves. The stenosis is caused by sterile ziegetations and cusp fusion.

6 513 Magilligan, Quinn, and Davila: Bacteremia, Endocarditis, and the Hancock Valve C Fig 3. Mitral valve from Patient 22, who had unsuspected candida on postmortem examination. The leaflets were intact, with nodular vegetations on the inflow side arranged in rows along the coapting D surfaces (A, B). The outflow (ventricular) side of the valve was completely clean (C, D). The sewing ring was well encapsulated in healthy scar tissue.

7 514 The Annals of Thoracic Surgery Vol 24 No 6 December 1977 successful treatment of I'VE. There was no evidence of valve dysfunction clinically or at postmortem examination. The fourth death was that of Patient 7, who had undergone mitral valve replacement 8 months after his episode of PVE and who died suddenly without evidence of valve dysfunction. Comment Patient Population Six of the 22 patients reported were drug addicts. The 27% incidence of addicts in a group of patients with active primary and secondary endocarditis is expected. The incidence of addicts among our total of 373 patients with Hancock valves is 12 (3.2%). Interestingly, the most frequent organism found in the addict group was Streptococcus, the most common infecting organism in endocarditis among nonaddicts [19]. The gram-negative and Streptococcus faecalis infections usually associated with drug abuse were more common in our nonaddict population. Active lnfective Endocarditis Since successful valve replacement in the face of active endocarditis was performed by Wallace [24] in 1965, a number of reviews have documented that valve replacement in the face of active infection may be undertaken with an operative mortality of 20 to 30% and a low late mortality [3, 14, 181. Surprisingly, infection of the prosthesis has not been a major factor. In a review of 293 patients who underwent valve replacement for endocarditis, the incidence of infection of the prosthesis was 4%. Among 162 patients with active infection, the incidence of PVE was 3% [141. This compares with a 0.5 to 1% incidence of early postoperative PVE in elective valve replacement [151. In collective reviews on the operative treatment of endocarditis, there are none in which tissue valves were routinely employed. The possible theoretical advantage of tissue valve prostheses in active infective endocarditis is not present in the stented porcine xenograft, because the sewing ring is similar to rigid prostheses and the seat of infection in I'VE is uniformly at the sewing ring-tissue annulus interface [21. Once infection occurs on a tissue valve prosthesis, however, cusp destruction usually follows whether the valve be fascia lata [ll, 211 or viable or nonviable homograft [6, 9, 23, 251, which leads to the conclusion that tissue valves are more vulnerable to infection [12]. In the present series of 9 patients who underwent 10 operations for acute infective endocarditis, there was no early infection of the xenograft. The 3 late infections at 3,12, and 32 months were from new foci, based on the length of the symptom-free interval, new organisms, or the introduction of a new source of infection. This is in agreement with the division of PVE into early and late cases based on whether it occurs within 2 months postoperatively. The former cases are invariably associated with septic events in the operative or postoperative period, whereas the latter are related to a source of infection distinct from the surgical procedure [41. We conclude, therefore, that the Hancock porcine xenograft, when inserted in the face of active infective endocarditis, is as resistant to infection as are rigid prostheses. Bacteremia As described by Sande 1201, bacteremias occuring within 24 days of valve replacement may not be associated with PVE, particularly if the organism is gram negative, if there is an extracardiac source of infection, and if there are no insufficiency murmurs. It cannot be assumed, however, that none of these patients has an infected prosthesis [261. Four of 5 patients reported by Dismukes and associates [81 who had gram-negative bacteremia in the early postoperative period had infection on the prosthesis. Clearly, any patient with bacteremia in the early postoperative period is at risk for PVE; in the present series the single patient with Staphylococcus aureus bacteremia was at great risk. Our other 5 patients fall into the group described by Sande [201. Three of the 6 patients were treated with long-term antibiotics, 2 received antibiotics for up to two weeks, and the patient with Bacteroides fvagilis bacteremia received no antibiotics because there was no longer a clinically septic course when the anaerobic cultures were reported. Thus we con-

8 515 Magilligan, Quinn, and Davila: Bacteremia, Endocarditis, and the Hancock Valve clude that the Hancock porcine xenograft is resistant to infection in the face of early postoperative bacteremias. Prosthetic Valve Endocarditis Prosthetic valve endocarditis occurring on a rigid prosthesis is associated with a mortality of 50 to 100% [13-16, 201. The mortality is highest when the infection occurs early (less than 60 days after valve replacement) [S, 16,22,27], and this is related to the more virulent organisms seen early, to postoperative system derangement, and possibly to altered host resistance inflicted by the pump oxygenator [lo]. The 11 patients in the present series all had late infections. Antibiotic therapy alone has been reported to be effective in eradicating infection in 37 to 44% of patients with late I'VE. In conjunction with operations for valve incompetence or persistent infection, antibiotic therapy has resulted in survivals of 58 to 64% [8, 221. The mortality in patients with I'VE is due to valve dysfunction, persistent infection, or embolus Among the 11 patients in the present series, Fig 4. Photomicrograph of the mitral valve from Patient 21, who had valvular stenosis produced by sterile vegetations. A cross-cut of one commissure included two leaflets on which are accumulated large vegetations. These consist of old and fresh thrombus; some portions show varying degrees of organization. (Masson trick rome; ~ 2.) 2 there were 3 deaths directly related to valve dysfunction and none to persistent infection or embolus. As evidenced by negative blood cultures or valve cultures, antibiotics were successful in curing infection in the 10 patients in whom they were employed. When compared to I'VE in rigid prostheses, the Hancock porcine xenograft appears to be easier to sterilize. The fact that the Hancock valve has proved easier to sterilize than rigid prostheses makes it similar to homografts [61. This is somewhat surprising because the sewing ring-tissue annulus interface is uniformly the site of infection in rigid I'VE [2]. The similarity in sterilization of the porcine xenograft and other tissue valves does not extend to durability in the face of I'VE. Infection in homograft valves, even if eradicated, almost universally results in cusp destruction that necessitates reoperation [l, 6,9,23,25]. Similarly, in late I'VE involving fascia lata or pericardial valves, 66% of survivors required reoperation because of valve incompetence [ll, 211. In the present series there were 3 valve failures. In 2 there was valve incompetence due to annular abscess, and 1 of these also had actual cusp destruction. In the third valve failure the cusps of the aortic and mitral xenografts were stiffened and obstructed by vegetations. Histology showed vegetations without organisms on calcified xenograft leaflets (Figs 4, 5).

9 516 The Annals of Thoracic Surgery Vol 24 No 6 December 1977 Fig 5. (Patient 21.) Higher power view of the area marked off in Figure 4. There is fragmentation of fibrinous material, and the horizontal layer of dark flecks beneath the vegetation represents calcification in the leaflet. (Masson trichrome; original magnification X2.50.) 1975 Our experience suggests, therefore, that in the face of PVE, the Hancock porcine xenograft is easier to sterilize than rigid prostheses and is more durable than other tissue valves. References 1. Angel1 WW, Buch WS, Shumway NE: The viable aortic homograft, Biologic Tissue in Heart Valve Replacement. Edited by MI Ionescu, DN Ross, GH Wooler. London, Butterworth, 1972, p Arnett EN, Roberts WC: Prosthetic valve endocarditis. Am J Cardiol38:281, Black S, O Rourke RA, Karliner JS: Role of surgery in the treatment of primary infective endocarditis. Am J Med 56:357, Block PC, DeSanctis RW, Weinberg AN, et al: Prosthetic valve endocarditis. J Thorac Cardiovasc Surg 60:540, Buch WS, Pipkin RD, Hancock WD, et al: Mitral valve replacement with Hancock stabilized glutaraldehyde valve. Arch Surg 110: 1408, Clarkson I M, Barratt-Boyes BG: Bacterial endocarditis following homograft replacement of the aortic valve. Circulation 43:987, Cohn LH, Sanders JH Jr, Collins JJ Jr: Aortic valve replacement with the Hancock porcine xenograft. Ann Thorac Surg 22:221, Dismukes WE, Karchmer AW, Buckley MJ, et al: Prosthetic valve endocarditis. Circulation 48:365, Gonzalez-Lavin L, Al-Janabi N, Ross DN: Longterm results after aortic valve replacement with preserved aortic homografts. Ann Thorac Surg 13:595, Hairston P, Lee WH Jr: Management of infected heart valves, Ann Thorac Surg 9:229, Ionescu MI, Abid A, Wooler GH: Experience with tissue heart valves, Recent Trends in Cardiovascular and Thoracic Surgery. Edited by JB Borman, Jerusalem, Alpha Press, 1975, p Ionescu MI, Pakraski BC, Holden MP, et al: Results of aortic valve replacement with frame supported fascia lata and pericardial grafts. J Thorac Cardiovasc Surg 64:340, Isom OW, Williams CD, Falk EA, et al: Long term evaluation of cloth covered metallic ball prostheses. J Thorac Cardiovasc Surg 64:354, Jung JY, Saab SB, Almond CH: The case for early surgical treatment of left-sided primary endocarditis. J Thorac Cardiovasc Surg 70:509, Killen DA, Collins HA, Koenig MG, et al: Prosthetic cardiac valves and bacterial endocarditis. Ann Thorac Surg 9:238, Madison J, Wang K, Gobel FL, et al: Prosthetic aortic valvular endocarditis. Cirsculation 51:940, 17. McIntosh CL, Michaelis LL, Morrow AG, et al: Atrioventricular valve replacement with porcine xenograft. Surgery 78:768, Parrott JCW, Hall JD, Keith WJ, et al: The surgical management of bacterial endocarditis. Ann Surg , Quinn EL, Madhaven T, Salahuddin N, et al: The increasing role of surgical therapy of bacterial endocarditis. Surg Clin North Am , Sande MA, Johnson WD Jr, Hook WK, et al: Sustained bacteremia in patients with prosthetic cardiac valves. N Engl J Med 286:1067, Senning A, Rothlin M: The late failure of autologous fascia lata valve grafts in the aortic position, Recent Trends in Cardiovascular and Thoracic Surgery. Edited by JB Borman. Jerusalem, Alpha Press, 1975, p Slaughter L, Morris JE, Starr A: Prosthetic valvular endocarditis. Circulation 47:1319, Trimble AS: Late results of homograft aortic valve replacement: a clinical and hemodynamic evaluation, Biologic Tissue in Heart Valve Replacement. Edited by MI Ionescu, DN Ross, GH Woller. London, Butterworth, 1972, p Wallace AG, Young GW, Osterhout S: Treatment of acute bacterial endocarditis by valve excision and replacement. Circulation 31:450, 1965

10 517 Magilligan, Quinn, and Davila: Bacteremia, Endocarditis, and the Hancock Valve 25. Wallace RB, Londe SP, Titus JL: Aortic valve replacement with preserved aortic.homografts. J Thorac Cardiovasc Surg 67:44, Weinstein L: Infected prosthetic valves: a diagnostic and therapeutic dilemma. N Engl J Med 286:1108, Wilson WR, Jammin PM, Danielson GK, et al: Prosthetic valve endocarditis. Ann Intern Med 82:751, Zuhdi N, Hawley W, VoehlV, et al: Porcine aortic valves as replacements for human heart valves. Ann Thorac Surg 17:479, 1974 Discussion DR. CHARLES R. HATCHER (Atlanta, GA): We are indebted to Dr. Magilligan and his associates for sharing this unique experience with us. I think everyone has been a little concerned about the possible performance of the Hancock valve in instances of infection. We were psychologically disarmed a bit by the information that antibiotic soaking could result in some leaflet damage, and deprived of that technique, I think our insecurity increased slightly. There was also a question of how the prosthesis would behave once we had active severe infection with a variety of organisms. Although this valve is currently being used in numerous centers, I think it takes a service such as Dr. Magilligan s to have the number of patients necessary to draw any meaningful conclusions about it. Without his experience with drug-addicted patients, none of us would have been able to study such a large number of infections; therefore, our conclusions would not have been so accurate. Since 1974, when we began to use the Hancock valve at Emory University, we have inserted 311 Hancock valves, with 2 patients developing endocarditis, 1 early and 1 late. Late endocarditis occurred in an elderly lady who had undergone mitral valve replacement; just over two years following valve insertion she returned with a Candida infection. This condition is extremely difficult to manage. We employed a technique that we had used successfully just once previously, in a patient who had a different valve. When the valve was removed, a huge vegetation was found projecting into the left atrium. Obviously, this type of infection is not responsive to any type of antibiotic program and necessitates replacement of the valve and reinsertion. In this situation, the concern about reinfection is great, and following the advice of Dr. Jerome Kay, all sutures and all infected material were removed, and a solution of amphotericin was used to fill the operative field. The chambers and the entire area in which the sutures would be placed were flooded with this solution. We then took a 30-minute break in the operation before we inserted the valve. There has been no active Candida infection in this patient in a year and a half following that maneuver. Thus, I suggest this method as a way of dealing with PVE. In conclusion, I believe our data support some of Dr. Magilligan s conclusions. I am not sure I would agree that we have established all of them, but these preliminary data are helpful in reassuring us about the behavior of the Hancockvalve in instances of early and late infection. DR. DENIS H. TYRAS (St Louis, MO): In 8 of 24 Hancock valves implanted during a 4-month period, cultures of the accompanying tissue remnants of aortic wall grew out a fastidious atypical mycobacterium. This organism which could be cultured only in thioglycolate broth, appeared at two to three weeks of incubation and could be subcultured only on a special thioglycolate agar incubated in a 5% carbon dioxide atmosphere. To determine the origin of the acid-fast organism, numerous cultures were made from sources that might have contaminated the valves or the thioglycolate medium. Saline, used for valve rinsing, and thioglycolate broth from sterility controls, including that from the same batches used for porcine valve culture but inoculated with cerebrospinal fluid, failed to grow acid-fast bacilli. Moreover, in 2 instances cultures of aortic wall coupons before rinsing grew out this atypical mycobacterium, as did the cotton balls and glutaraldehyde solution that accompanied the valve. The identity of this atypical mycobacterium is still under investigation. Its growth characteristics are most typical of Mycobacterium bovis, known to be pathogenic in swine. However, the Center for Disease Control in Atlanta has informed us that a porcine valve culture from another institution has grown M. chelonei, a member of Runyon s group IV. There have been 2 reports of endocarditis resulting from this organism. Our patients are being followed carefully, and only 1 has shown any problems to date. Five months after mitral valve replacement this patient developed a large pericardial effusion which, at the time of pericardiectomy, was found to contain acid-fast organisms. We are not certain of the clinical importance of this finding. Collins and Montalbine have shown that buffered 0.2% glutaraldehyde is not mycobactericidal, and we believe that for a valve to be labeled sterile it is vital to show evidence of negative cultures in thioglycolate broth under anaerobic conditions for three weeks, so as to assure that the valves are free of this organism before the manufacturer forwards them to individual users. We are currently evaluating the effects on this organism of higher concentrations of glutaraldehyde such as are used with the Carpentier-Edwards porcine heterograft. DR. AGUSTIN ARBULU (Detroit, MI): The authors mention that 1 patient required reoperation because of

11 518 The Annals of Thoracic Surgery Vol 24 No 6 December 1977 valve destruction and that they found a subannular aneurysm. We have found these subannular aneurysms to be localized mainly below the right and noncoronary cusps. To date we have seen 18 patients with bacterial endocarditis. The bacteria were gram positive. Nine of these patients were heroin addicts. In our initial experience we repaired the subannular aneurysm with sutures, and in more than half of our patients the aortic insufficiency recurred. Then we modified our technique, and we considered it extremely important to repair the subannular aneurysm with a patch so as to afford support to the prosthesis. We have repaired 8 subannular aneurysms by the patch technique and have had no complications. All these patients were cured of their aortic insufficiency. I wonder if Dr. Magilligan will comment on this problem. My recommendation is that a subannular aneurysm should be carefully repaired with a patch prior to insertion of the prosthesis. DR. WILLIAM ANGELL (San Jose, CA): We have been concerned with heterograft valves and postoperative infection. We have also wondered whether a perivalvular leak did not precede infection in a number of patients. I wonder if the authors could indicate the number of valves in this experience that had associated perivalvular leaks. DR. SAMUEL L. KALUSH (Milwaukee, WI): Dr. Magilligan, were these patients maintained on long-term antibiotics? DR. MAGILLIGAN: We agree with Dr. Hatcher that the appearance of candida on a prosthetic valve is indication for removal. We, too, have soaked the annulus with amphotericin, but we have not soaked the heart with it because we have been told that there is a possibility it may be toxic to the myocardium. I was very interested in Dr. Tyras report of atypical mycobacteria. We became interested in this when we investigated 2 patients with tuberculosis following valve replacement. These were typical cases of human Mycobacteriurn tuberculosis, but in talking to veterinarians at Michigan State University, we found an expert on swine TB who told us that about 70,000 swine are retained in federally inspected slaughter plants each year because of Runyon group I11 organisms. We have not found any atypical mycobacteria. Although what Dr. Tyras has told us is important, I don t think it should throw us into a panic about using the valve, because the incidence of infection in humans, to my knowledge, has been zero. We are indebted to Dr. Arbulu for his description of how to repair subannular aneurysms. This is a terrible problem, and in our patient who did not survive operation, we attempted to repair the aneurysm with a patch and a new prosthesis. Bleeding, however, became an overwhelming problem. The 2 aortic valves that we showed here had perivalvular leaks. There was the kind of dehiscence that is seen in a rigid-frame prosthesis involved with endocarditis. Preexisting perivalvular leak was present in 1 other patient in whom the mitral valve had been improperly seated at the original operation. In PVE involving a rigid-frame prosthesis, I think the infection occurs at the tissue annulus-sewing ring interface in l0oo/o of the patients, and this is where dehiscence will occur. The role of glutaraldehyde in dehiscence is not clear, but we have seen dehiscence in only 2 patients, each of whom had endocarditis. With regard to long-term antibiotic therapy (patients with the early bacteremias), 3 had long-term therapy and 3 did not. This reflects our uncertainty over whether this is really infection of the valve. Among the patients with acute infective endocarditis who had valves replaced during active infection, all had long-term antibiotics following operation except for 1 who was treated for only two weeks. Ten of 11 patients with I VE received appropriate antibiotics. We continued this, when possible, for six weeks.