Clinical durability of the Hancock porcine bioprosthetic valve

J

THoRAc CARDIOVASC SURG

80:824-833, 1980

Original Communications

Clinical durability of the Hancock porcine bioprosthetic valve The principal feature of the Hancock xenograft bioprosthesis which remains to be completely defined is long-term durability. This report provides extended data regarding valve durability derived from a data base of 1,407 patients (707 aortic [AVR] and 700 mitral [MVR] replacements) who received Hancock bioprostheses between 1971 and 1979; cumulative duration offollow-up was 1,732 patient-years for AVR and 1.,843for MVR patients, with a maximum follow-up duration of 8.4 years. One hundred seventy-nine patients were followed for more than 5 years and 67 for more than 6 years. Valve failure was defined on the basis of one or more of the following criteria: (I) postoperative development of a new regurgitant murmur, (2) thrombotic valvular occlusion, (3) infective endocarditis resulting in reoperation or death, and (4) hemodynamic valvular dysfunction confirmed by catheterization and resulting in reoperation or death. Twenty-one such failures occurred among all AVR patients and 23 among all MVR patients. The actuarial probability offreedom from valve failure (all causes) was 95.4% ± 1.2% (±SEM) for adult AVR patients 5 years postoperatively and 90.9% ± 2.6% for adult MVR patients 6 years postoperatively. The probability offreedom from primary tissue failure in adults was 99% ± I % in AVR patients at 5 years and 94.3 % ± 2.4% in MVR patients at 6 years. The linearized incidence of primary tissue failure in children «15 years old) was 9.8% per patient-year (combined AVR and MVR patients), compared to 0.2% per patient-year among all adult patients in the analysis. The combined actuarial incidence of primary tissue failure among adults with AVR and MVR was 98.6% ± 0.7% at 5 years and 94.2% ± 2.3% at 6 years; thus there appears to be a slight acceleration in the rate of valve tissue failure between 5 and 6 years after operation. The incidence offailure, however, remains acceptably low through 6 years offollow-up , and continued clinical use of the xenograft bioprosthesis seems warranted.

Philip E. Oyer, M.D. (by invitation), D. Craig Miller, M.D. (by invitation), Edward B. Stinson, M.D. (by invitation), Bruce A. Reitz, M.D. (by invitation), Ricardo J. Moreno-Cabral, M.D. (by invitation), and Norman E. Shumway, M.D., Stanford, Calif.

T

he performance of the glutaraldehyde-fixed xenograft bioprosthetic valve has now been well characterized in terms of thrombogenicity and susceptibility to infection. r. 2 Furthermore, several authors have docFrom the Department of Cardiovascular Surgery, Stanford University School of Medicine, Stanford, Calif. Read at the Sixtieth Annual Meeting of The American Association for Thoracic Surgery, San Francisco, Calif., April 28 to 30, 1980. Address for reprints: Philip E. Oyer, M.D., Ph.D., Department of Cardiovascular Surgery, Stanford University School of Medicine, Stanford, Calif. 94305.

824

umented generally satisfactory hemodynamic characteristics, although unacceptably large transvalvular pressure gradients have occasionally been noted, particularly across the smaller-sized aortic bioprostheses.>:" Thus the principal feature of the xenobioprosthesis which remains to be elucidated is its clinical durability, a topic of substantial importance inasmuch as bioprosthetic valves are being implanted on an increasingly widespread basis. The present analysis extends our previously reported experience with the Hancock xenobioprosthetic valve, with particular emphasis on characterization of its clinical durability.

0022-5223/80/120824+ 10$01.00/0 © 1980 The C. V. Mosby Co.

Volume 80

Hancock porcine bioprosthetic valve

Number 6

825

December, 1980

Patients and methods Data base. All patients who have undergone aortic (AVR) or mitral (MVR) valve replacement with the Hancock xenobioprosthesis on the University Cardiovascular Surgery Service, Stanford University Medical Center, since 1971 were included in this analysis. Prior to 1974, such valves were used on an individual basis. Since October, 1974, however, all patients undergoing cardiac valve replacements have received bioprostheses. The analysis includes 707 AVR patients and 700 MVR patients. Since November, 1976, 209 AVR patients whose aortic annular diameter was 25 mm or less received modified orifice bioprostheses. The distribution of patients in this analysis according to age at operation is given in Fig. 1. Operative technique. Procedures for valvular implantation have remained constant throughout the study period. After thorough debridement of annular calcification, mattress sutures of braided polyester were used to fix the bioprosthesis to the anulus. Teflon felt pledgets were used only in cases of severe annular friability. Myocardial preservation was achieved with topical hypothermia and moderate systemic hypothermia combined with low cardiopulmonary bypass flow rates (35 to 45 mIlkg/min); more recently, cold (4° C) hyperkalemic cardioplegia has been employed additionally in some patients. Anticoagulation. Currently, warfarin sodium anticoagulation is given to all A VR patients for 6 weeks and MVR patients for 12 weeks after operation. Patients whose anatomic findings at the time of operation suggested an increased potential for late development of thromboembolic complications were given anticoagulation therapy indefinitely. In addition, long-term anticoagulation was usually instituted in patients who had a postoperative systemic arterial embolus. Follow-up data. All patients or their primary physicians, or both, were contacted by telephone during a 3 month closing interval, at which time pertinent features of the clinical status of each patient were assessed. The distribution of discharged patients as a function of length of follow-up is shown in Fig. 2. Sufficient numbers of patients for meaningful statistical analyses were available 5 years postoperatively for A VR patients and 6 years postoperatively for MVR patients. The cumulative duration of follow-up for all A VR patients was 1,732 patient-years and that for all MVR patients, 1,843 patient-years. The cumulative follow-up duration for A VR patients less than 15 years old was 9 patientyears (n = 3) and that for MVR patients, 42 patientyears (n = 9). Overall, 12 AVR patients (1.7%) and six MVR patients (0.9%) were lost to follow-up.

250

232

AORTIC VALVE REPLACEMENT tm MITRAL VALVE REPLACEMENT

0 (/)

I-

Z

200

w j::::

« Q.

214

.;.~

',",

157 ..:

150

~··i

u..

1 11

0

a: 100 fw

D:l

:!E

::>

z

59

50 0

~ <15

fbi

f\i

W

<

.~

78

i~

'>J:.

:~.

t~.g

ft

~f

Z~·

)~

~.... ~:'

iU

~

;;.:

~~.

:~:

:),i I;. {i.

*~-

,~

',~~ l:" ' :S;'

~

>, ~! 15-24 25-34 35-44 45-54 55-64

;1:

~42

:-:: >.;;

.;-

".

>65

PATIENT AGE AT OPERATION (years)

Fig. 1. Distribution of patients in the analysis according to the age at the time of operation. Data include only discharged patients.

Definition of valve failure. Criteria for the diagnosis of valve failure remain unchanged from those defined in previous reports from this institution.' One or more of the following circumstances were considered to be indicative of bioprosthesis failure: (1) postoperative development of a new regurgitant murmur, unless proved to be peri prosthetic in origin, (2) multiple embolic episodes or thrombotic occlusion of the valve necessitating reoperation or resulting in death, (3) infective endocarditis resulting in reoperation or death, and (4) confirmed hemodynamic valvular dysfunction sufficient to necessitate reoperation or cause death. "Primary tissue failure" was diagnosed either at reoperation or at postmortem examination when the valve leaflets exhibited gross calcification and/or perforations or disruption in the absence of antecedent endocarditis. Patients who exhibited large trans valvular gradients and required replacement of the bioprosthesis, but who appeared to have full valve leaflet mobility and integrity at reoperation, were considered to have valve failure; such failure was labeled "intrinsic stenosis. " Data analysis. Actuarial analysis of clinical valve failure was performed according to the nonparametric method of Kaplan and Meier. 7 The Gehan" modification of the Wilcoxon test was used to assess the significance of differences in actuarial rates so derived. Actuarial data points were plotted only if associated standard errors of the mean (SE) were reasonably small « ±5%), since data points with larger standard errors are of little practical utility and, moreover, may be misleading. The maximum likelihood ratio test" or COX'SI0 F-test was used to test the significance of observed differences in occurrence rates of certain events.

The Journal of

Oyer et al.

826

Thoracic and Cardiovascular Surgery

200 D AORTIC VALVE REPLACEMENT t:lJ MITRAL VALVE REPLACEMENT

(I)

I-

z 150 w

I- 146~0 "":":.:

jj I··.~d\ .

I-

«e,

u.. 0

100

~

:::l

50

~~~'.

~~~

~ ~.

\~:

I-

":,

y.

Z

':.

.~;,

~{~;~

0

f} 0-1

-::. ~.;

~~

~

w

W 1-2

125 103

~

.i~

ex:

co

I 10

:~:

:i#:

I-

1~9

122

~3

~

i~

.•..

:':1

:~

~;~

~;1

:;,.

:,:'

f·; OJ

~:

.;".

2-3

3-4

-

95

62

~

52

~ ~l

60

1j, . .:.~

,:.~;

~,: .

~MJ:.•

4-5

"

5-6

-

53

rr.. 'i!:

}~

-

9

~;

.!..m ,I.

6-7

7-8

3

I

8-9

YEARS POSTOPERATIVE

Fig. 2. Distribution of patients in the analysis according to length of postoperative follow-up. Only discharged patients are included.

85

o AVR • MVR

[n = [n =

2

704) 691)

3

4

5

6

YEARS POSTOP

Fig. 3. Actuarial incidence of bioprosthetic valve failure from years old). Brackall causes among adult patients (~15 ets = 1 standard error of the mean. AVR, Aortic valve replacement. MVR, Mitral valve replacement.

Chi square contingency analysis or the Fisher exact test was used to compare discontinuous variables. Cox's 11 multivariate regression analysis for censored data was used to examine correlations between certain independent variables and bioprosthesis durability.

Results Overall bioprosthesis failure. The actuarial incidence of aortic and mitral valve failure from all causes among adult patients is shown in Fig. 3. Although the maximum follow-up interval extends beyond 8 years, data are shown only through 5 and 6 years for AVR and MVR patients, respectively, since the number of patients followed for longer periods in each group was too small to allow meaningful conclusions. The linearized rate of overall bioprosthesis failure, 1.1 % per patientyear, was the same for AVR and MVR patients. Al-

though the incidence of valve failure was identical for AVR and M VR patients within the constraints of the follow-up period, the proportionate contributions of various modes of valve failure among adult patients differed according to valve location, as shown in Fig. 4. Thus, among AVR patients, endocarditis resulting in reoperation or death accounted for the majority of failures; in contradistinction, among MVR patients, primary tissue failure was the most prevalent failure mode. This difference was not totally attributable to the slightly longer follow-up duration in the MVR group.

Primary tissue failure (impact of patient age). Examination of the impact of multiple preoperative variables on valve durability using Cox's multivariate regression and covariance analysis confirmed (p < 0.01) previous observations that the incidence of valve failure is higher in children than adults. 12-15 The threshold age above which the risk of valve failure appeared to plateau was 15 years. The mode of failure in patients less than 15 years old (children) was primary tissue failure in all cases; in each case severe leaflet fibrosis and calcification which resulted in stenosis was apparent. In some cases regurgitation resulting from leaflet perforation at the margins of heavily calcified areas was also present. In this age group, two of three aortic valves have failed within 3 years, and three of nine mitral valves have failed within 4 years after implantation. Thus the combined A VR and MVR linearized rate of bioprosthesis failure among children is 9.8% per patient-year. This is in distinct contrast to the combined A VR and MVR linearized rate of primary tissue failure of 0.25% per patient-year in adults (0.12% per patient-year for A VR and 0.39% per patient-year for MVR patients). Comparison of these rates of primary tissue failure in adults and children by

Volume 80

Hancock porcine bioprosthetic valve

Number 6

827

December. 1980

AORTIC (n = 19)

MITRAL (n = 20)

PRIMARY TISSUE FAILURE INTRINSIC 10.5% (2 pts) STENOSIS 5.3% (1 pt)

PRIMARY TISSUE FAILURE 35% (1 pts]

THROMBOSIS 5.3% (1 pt) REGURGITANT MURMUR 15.8% (3 pts]

ENOOCARDITIS 63.2% (12 pts]

INTRINSIC STENOSIS 5% (1 pt]

THROMBOSIS 10% (2 pts]

REGURGITANT MURMUR 25% (5 pts)

Fig. 4. Proportionate contributions of the various modes of bioprosthesis failure to the overall number of valve failures observed among adult A VR and MVR patients.

means of the maximum likelihood ratio test (which takes into account differences in cumulative follow-up duration between groups) confirmed the significance of these differences (p < 0.005). Actuarial analysis was not appropriate to express valve failure rates in children because the small number of patients in this group resulted in large standard errors associated with each data point, so that meaningful conclusions were impossible. The incidence of primary tissue failure among adult AVR and MVR patients, calculated by the actuarial method, is shown in Fig. 5, A. The probability of freedom from primary tissue failure for AVR patients at 5 years was 99% ± 1% (±SE) and that for MVR patients, 94% ± 2.4% at 6 years. The incidence of tissue failure remains linear through the initial 5 years after operation for both AVR and MVR patients. Thereafter, an increase in the rate of tissue failure in MVR patients can be appreciated. Three instances of primary failure occurred among MVR patients between 5 and 6 years postoperatively and accounted for the increase in the rate ofbioprosthesis failure after 5 years. Thus, through the first 5 years after operation, the rate of tissue failure in adult patients was 0.24% ± 0.12% (SE) per patient-year; however, between 5 and 6 years after operation the rate was 2.3% ± 1.3% per patient-year. The difference is statistically significant (p < 0.01) as assessed by Cox's F-test. The combined actuarial incidence of primary tissue failure among all 1,395 adult AVR and MVR patients is shown in Fig. 5, B for comparison with similar data reported by others. 16 The late acceleration in the rate of tissue failure is again evident in this combined analysis; assessed in this way, the probability of freedom from primary tissue failure 6 years after operation was 94% ± 2.3% (SE). The rate of tissue failure for the combined group of 1,395 adult patients during the initial 5 postoperative years (0.15% ± 0.07% per patient-

year) differs significantly (p < 0.00 l) from that observed between 5 and 6 years after operation (2.5% ± 1.3% per patient-year). Among the 14 instances of primary tissue failure in 1,407 A VR and MVR patients of all ages, only one case of leaflet perforation unassociated with calcification was noted (1.8 years postoperatively), possibly an instance of random tissue failure rather than a manifestation of an ongoing degenerative process. There have been no instances of primary tissue failure among adult patients who received modified orifice aortic bioprostheses, although a modified orifice valve in a lO-year-old patient has failed because of leaflet calcification, and one adult with a modified orifice valve has a regurgitant murmur of unknown origin. Endocarditis. The impact of infectious complications on bioprosthetic valve durability has not been fully delineated, in part because such complications are relatively infrequent occurrences. A total of 21 cases of infected aortic bioprostheses and eight cases of infected mitral bioprostheses have been identified in this patient series; thus the linearized occurrence rates for endocarditis involving bioprostheses are 1.2% and 0.4% per patient-year for A VR and MVR patients, respectively. Among A VR patients, 12 received medical therapy only; nine were cured and survived while three died. Nine A VR patients required reoperation, usually because of either persistent evidence of infection (despite antibiotic treatment) or congestive heart failure resulting from valvular or perivalvular regurgitation; three of these patients were thereby cured of their infections while six died of persistent infection or complications thereof. Among the eight MVR patients with infective endocarditis, three were cured by medical therapy alone. One died without treatment, since the appropriate diagnosis was not established before death. Four MVR pa-

The Journal of

828

Oyer et at.

Thoracic and Cardiovascular Surgery

o AVR (n • MVR (n

= 704) = 691)

2

3

4

5

6

r 7

YEARS POSTOP

1

.. 1

0

~ 2

4

COMBINED AVR AND MVR PATIENTS

2

3

4

5

6

7

YEARS POSTOP

Fig. 5. A. Actuarial incidenceof primary tissue failure among adult patients (;::,15 years old). Separate analyses for AVR and MVR patients are shown. B. Combined actuarial incidence of primary tissue failure among adult AVR and MVR patients. The number of such failures that occurred during each yearly postoperative interval are shown. Brackets = I standard error of the mean. tients required reoperation; two were cured and two died of complications of the infection. Thus 17 of 28 patients with endocarditis diagnosed before death (61 %) have survived infections involving bioprosthetic valves by virtue of medical or combined medical and surgical treatment and, thereafter, have experienced no identifiable valvular dysfunction. Regurgitant murmur. A total of eight patients (five MVR and three A VR) were considered to have bioprosthetic valve failure on the basis of development of a new regurgitant murmur of unknown origin between 4 months and 5 years after operation. Their clinical status has not warranted recatheterization or reoperation; all are in New York Heart Association Functional Class I or II. Inasmuch as a proportion of these patients may have perivalvular leaks rather than valvular regurgitation, some may not have actual valve leaflet degeneration. Thrombosis and thromboemboli. Three instances of thrombosis of bioprostheses (one A VR and two MVR) have occurred. One patient, receiving appar-

ently adequate anticoagulation, died of acute thrombosis of a mitral prosthesis 2 weeks after discharge from the hospital following operation. No predisposing causes for thrombosis were identified. The remaining two patients manifested symptoms of valvular stenosis between 1.5 and 2 years after their initial operations and underwent subsequent successful reoperation. Findings at reoperation consisted of acute and chronic thrombotic deposits on thickened and fibrotic valve leaflets in both patients. The thrombosed aortic valve evidenced inward deformation of the support struts, possibly as a consequence of polypropylene "creep," which may have contributed to the deposition of fibrinous material on the outflow surface of the valve . No patient in this series has required reoperation as a result of multiple thromboembolic episodes. The incidence of thromboemboli after bioprosthetic valve replacement has not changed substantially from that stated in earlier follow-up reports. I Influence of patient-associated variables on valve durability. The impact of various patient-associated variables on bioprosthesis durability were examined by univariate discriminant analysis. The variables examined included cause of the pre-existing valvular disease, preoperative pulmonary, hepatic, and renal disease, hypertension, diabetes mellitus, hyperlipidemia, age, sex, cardiac rhythm, and various hemodynamic variables obtained at preoperative catheterization. Significant (p < 0.05) associations between patient age at operation, preoperative renal dysfunction, and preoperative pulmonary disease and valve durability were found. Multivariate discriminant analysis, however, subsequently revealed that of these three parameters, only patient age at operation was independently correlated with bioprosthesis durability, as previously outlined. Mortality rate of reoperation for valve failure. Thirteen patients (nine A VR and four MVR) underwent reoperation for valve failure resulting from endocarditis. Eight operative deaths occurred, for an overall operative mortality rate of 62% in the presence of active bioprosthesis infection. In contrast, all 13 patients (three A VR and 10 MVR) who underwent reoperation for bioprosthetic tissue failure in the absence of infection survived replacement of the bioprosthesis. The difference in operative mortality rates for those reoperated upon for endocarditis compared to primary tissue failure (62% versus 0%) is significant (p = 0.003). Discussion Degenerative changes in xenobioprosthetic valves have been shown by ultrastructural analysis to com-

Volume 80 Number 6 December, 1980

mence shortly after implantation and progress through several stages to culminate in severe collagen disruption as early as 2 years thereafter.!" Simultaneous regenerative processes, however, have not been observed, so that the inescapable conclusion is that such valves eventually will fail. Functional lifetime, therefore, remains the single most important performance feature of bioprosthetic valves yet to be established. This update of previous reports from our institution allows firmly based conclusions regarding the durability of the Hancock bioprosthesis through 5 and 6 years for valves in the aortic and mitral positions, respectively. Although patients have been followed for longer intervals, they number too few to provide statistically meaningful data. The criteria used to define overall valve failure were designed to describe equally well all major types of failure that may occur in both bioprosthetic and mechanical valves. Thus the use of such general criteria to define valve failure allows valid comparison of the performance of these two types of prosthesis, even though the specific modes of failure observed for one type of valve may differ from those of another. The proportion of adults free from overall failure of the Hancock aortic bioprosthesis (95.4% at 5 years) and the mitral bioprosthesis (90.9% at 6 years) compare favorably with similarly defined failure rates associated with certain mechanical valves. IS. Definition of the age-related dependency of bioprosthesis failure rates is an important issue. Appropriate decisions regarding the use of tissue valves in children require an estimate of the probability of early failure, balanced against the clear advantages of bioprostheses in this age group, as outlined by Geha and associates. 13 As a corollary, the relatively high failure rates in children should not influence the decision to use such valves in adult patients. For these reasons, valve failure data for each patient group must be analyzed separately. Such data, as reported herein, allow reasonably certain conclusions regarding bioprosthesis durability in adult patients. Although children clearly exhibit a higher risk of valve failure than adults,12-15 the magnitude of this risk cannot yet be determined accurately because of the small numbers of young patients available for analysis; thus the suitability of bioprostheses for use in children cannot be assessed on firm statistical grounds at this time. The observed rate of primary tissue failure in adults between 5 and 6 years postoperatively was approximately tenfold greater than during the initial 5 postoperative years of follow-up (2.3% ± 1.3% versus 0.24% ± 0.12% per patient-year for MVR patients and 2.5% ± 1.3% versus 0.15% ± 0.07% per patient-year

Hancock porcine bioprosthetic valve

829

for the combined group). This late acceleration in the rate of tissue failure was found to be statistically significant, which suggests that the observed increase was not due to chance alone. Nevertheless, the cumulative follow-up duration for patients at risk for more than 5 years after valve replacement was small (133 patientyears for MVR patients and 157 patient-years for the combined A VR and MVR patient group); thus there were relatively large standard errors associated with the observed failure rates during this late postoperative interval. Therefore, the true magnitude of the increase in bioprosthetic tissue failure rates beyond 5 years after operation cannot yet be established with certainty. The incidence of bioprosthetic valve failure recently reported by Magilligan and associates!" in a large group of patients observed for an extended period of time closely parallels that observed in this series. Because the definition of valve failure used in that report was similar to our definition of "primary tissue failure," valid comparisons can be made. The slightly higher actuarial probability of freedom from bioprosthetic tissue failure at 6 years for the combined group of A VR and MVR patients observed in our series, compared to that reported by Magilligan (94.2% ± 2.3% SE versus 90.8% ± 2.4% SE), presumably reflects the inclusion of both adults and children in the latter analysis. We have not been able to confirm a statistically significant increased risk of tissue failure in patients between 15 and 35 years of age, as reported by Magilligan. In fact, no instance of primary tissue failure has occurred in any patient between 15 and 34 years of age in our series. Polypropylene "creep" involving the supporting struts of Hancock valves has been reported to result in valve failure necessitating reoperation.:": 20 Only one instance of possible polypropylene creep, with inward deformation of the struts resulting in partial late thrombosis of an aortic prosthesis in an adult, has been observed in this series of 1,407 patients. This patient exhibited symptoms of aortic stenosis and underwent successful replacement of the bioprosthesis. Thus the incidence of Hancock valve failure attributable to polypropylene strut deformation is exceedingly small. Among the 13 patients in this series who underwent reoperation for primary tissue failure, there were no operative deaths. This favorable outcome of reoperation undoubtedly results in part from the fact that catastrophic valve failures were not observed. Rather, the relatively slow onset of symptoms of valvular dysfunction allowed an orderly approach to bioprosthesis replacement in all cases. Continuation of low reoperative mortality rates, however, requires assiduous postoperative follow-up to ensure that reoperation is performed in a timely manner, before significant bio-

The Journal of

830

Oyer et al.

prosthesis dysfunction results in irreversible myocardial damage. 21 Regular evaluations will be particularly important for patients with valves implanted for more than 5 years, since bioprosthetic tissue failure rates begin to increase after this interval. Echocardiographic'" or gated computerized tomographic methods for assessing progressive structural changes in bioprosthetic valves are under development and eventually may play an important role in long-term patient follow-up. Infective endocarditis involving bioprosthetic valves is associated with a substantial risk of death, as it is with mechanical prostheses. Thus the overall mortality rate in this series of patients treated with medical or combined medical and surgical therapy was 39%. This rate is almost identical to that reported earlier from this institution for endocarditis involving Starr-Edwards mechanical prostheses. 23 Moreover, the incidence of bioprosthetic valve infections does not differ substantially from that reported for mechanical valves.P: 24 Thus bioprosthetic valve failure resulting from endocarditis does not appear to be associated with greater morbidity or mortality than do similar complications involving mechanical prostheses. A recently reported policy" of discontinuation of use of bioprostheses in the mitral position, based primarily on results of hemodynamic studies performed more than 5 years after operation, can be criticized because of the small number of patients studied and the selective nature of the patient population. Moreover, although the hemodynamic data obtained in one third of the patients studied suggested obstruction to flow, the rate of progression of hydraulic dysfunction (and therefore the implications for appropriate timing of reoperation) could not be predicted. Summary

Although the incidence of clinically important bioprosthetic tissue failure through 6 years of follow-up remains small, the apparent increase in failure rates between 5 and 6 years after implantation, as observed in both this series and that from Detroit," mandates continued close surveillance of patients with xenograft bioprostheses in order to assure timely reoperation when required. Available data, however, continue to support the clinical use of such valve substitutes in adult patients, particularly inasmuch as elective replacement of such valves for primary tissue failure appears to carry a low operative risk. The rate of bioprosthesis failure among children is clearly higher than that observed in adult patients; how-

Thoracic and Cardiovascular Surgery

ever, conclusive quantification of the time-related risk for young patients is not yet possible on the basis of existing data. Therefore, the advisability of bioprosthesis implantation in children remains to be determined. We would like to acknowledge the contributions of Voy Wiederhold and Marie Hu in regard to statistical techniques and computer programming support used in assessing the data contained herein. Karen Mueller was instrumental in compiling patient follow-up information. REFERENCES

2

3

4

5

6

7

8

9

10 11 12

13

Oyer PE, Stinson EB, Reitz BA, Miller DC, Rossiter SJ, Shumway NE: Long-term evaluation of the porcine xenograft bioprosthesis. J THORAC CARDIOVASC SURG 78:343-350, 1979 Cohn LH, Koster JK, Mee RB, Collins JJ: Long-term experience with porcine aortic valve xenografts. Circulation 60:Suppl 3:87-92, 1979 Rossiter SJ, Miller DC, Stinson EB, Oyer PE, Reitz BA, Moreno-Cabral RJ, Mace JG, Robert EW, Tsagaris TJ, Sutton RB, Alderman EL, Shumway NE: Hemodynamic and clinical comparison of the Hancock modified orifice and standard orifice bioprostheses in the aortic position. J THORAC CARDIOVASC SURG 80:54-60, 1980 Lurie AJ, Miller RR, Maxwell KS, Grehl TM, Vismara LA, Hurley EJ, Mason DT: Hemodynamic assessment of the glutaraldehyde-preserved porcine heterograft in the aortic and mitral position. Circulation 56:Suppl 2: I04110, 1977 Craver JM, King SB, Douglas JS, Franck RH, Jones EL, Morris DC, Kopchak J, Hatcher CR: Late hemodynamic evaluation of the Hancock modified-orifice aortic bioprosthesis. Circulation 6O:Suppl 1:93-97, 1979 Ubago JL, Figueroa A, Colman T, Ochoteco A, Duran C: Hemodynamic factors that affect the calculated orifice areas in the mitral Hancock xenograft valve. Circulation 61:388-394, 1980 Kaplan EL, Meier P: Non-parametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958 Gehan EA: Generalized Wilcoxon test for comparing arbitrary singly-censored samples. Biometrika 52:203-223, 1965 Gross AJ, Clark VA: Survival distributions. Reliability applications in the biomedical sciences, New York, John Wiley & Sons, Inc., pp 240-243 Cox DR: Some sample tests for Poisson variates. Biometrika 40:354-360, 1953 Cox DR: Regression models and life tables. J R Stat Soc (Ser. B) 34:187-220, 1972 Kutsche LM, Oyer PE, Shumway NE, Baum D: An important complication of Hancock mitral valve replacement in children. Circulation 60:Suppl 1:98-103, 1979 Geha AS, Laks H, Stansel HC Jr, Cornhill JF, Kilman

Volume 80

Hancock porcine bioprosthetic valve

Number 6 December. 1980

14

15

16

17

18

19

20

21

22

23

24 25

JW, Buckley MJ, Roberts We: Late failure of porcine valve heterografts in children. J THORAC CARDIOVASC SURG 78:351-364, 1979 Bachet J, Bical 0, Goudot B, Menu P, Richard T, Barbagelatta M, Guilmet 0: Early structural failure of porcine xenografts in young patients, Bioprosthetic Valves, F Sebening, ed., Munich, 1979, Deutsches Herzzentrum, p 341 Sanders SP, Freed MD, Norwood WI, Castaneda AR, Nadas S: Early failure of porcine valves implanted in children (Abstr). Am J Cardiol 45:449, 1980 Magilligan OJ, Lewis JW, Jara FM, Lee MW, Alam M, Riddle JM, Stein PO: Spontaneous degeneration of porcine bioprosthetic valves. Ann Thorac Surg (in press) Ferrans VJ, Spray TL, Billingham ME, Roberts We: Structural changes in glutaraldehyde-treated porcine heterografts used as substitute cardiac valves. Am J Cardiol 41:1159-1184, 1978 Oyer PE, Griepp RB, Stinson EB, Shumway NE: Valve replacement with the Starr-Edwards and Hancock prostheses. Comparative analysis of late morbidity and mortality. Ann Surg 186:301-309, 1977 Salomon NW, Copeland JG, Goldman S, Larson OF: An unusual complication of the Hancock porcine heterograft. Strut compression in the aortic root. J THORAC CAROlOVASC SURG 77:294-296, 1979 Magilligan OJ, Lewis J, Jara FM, Lakier JB, Davila JC: Indications for reoperation after bioprosthetic valve implantation, Bioprosthetic Valves, F Sebening, ed., Munich, 1979, Deutsches Herzzentrum, p 319 Rossiter SJ, Miller DC, Stinson EB, Oyer PE, Reitz BA, Shumway NE: Aortic and mitral reoperations. Early and late results. Arch Surg 114:1279-1283, 1979 Alam M, Madrazo A, Magilligan OJ, Goldstein S: M-mode and two dimensional echocardiographic features of porcine valve dysfunction. Am J Cardiol 43:502-509, 1979 Rossiter SJ, Stinson EB, Oyer PE, Miller DC, Schapira J, Margin RP, Shumway NE: Prosthetic valve endocarditis. J THoRAc CARDIOVASC SURG 76:795-803, 1978 Stinson EB: Surgical treatment of infective endocarditis. Prog Cardiovasc Dis 22: 145-168, 1979 Lipson L, Kent KM, Mcintosh CL, Rosing DR, Bonow RO, Condit J, Epstein S, Morrow AG: Hemodynamic evaluation of porcine heterografts in the mitral position for more than five years (Abstr). Am J Cardiol 45:486, 1980

Discussion DR. JOHN KENNEDY London. England

I would like to compliment Dr. Oyer and his colleagues for their interesting study on the long-term durability of xenograft valves and to focus my remarks on the area of primary tissue failure.

83 1

As in mechanical prostheses, the quality of a bioprosthesis depends upon the material properties. A mammalian aortic valve is a compound biopolymer with collagen, elastin, and mucopolysaccharide elements. In the case of stored human aortas at the National Heart Hospital, London, Drs. Arridge, Olsen, and I have found random variations in the mechanical properties and histochemical deterioration in the elastin and mucopolysaccharide elements after 16 days of storage. Much has been said about the effect of chemical fixation on the collagen properties of xenograft valves. Do the authors have any comments about the other important elastin and mucopolysaccharide elements in their processed valves that might affect long-term durability? DR. CARLOS G. DURAN Santander. Spain

We have been encouraged by the results presented by Dr. Oyer to continue using the Hancock bioprosthesis. Between June, 1974, and June, 1979, we have used 734 Hancock valves with a maximum follow-up of 5.5 years, minimum of 6 months, and average of 2.7. There have been 12 instances of primary valve failure: aortic rupture, three cases (two secondary to bacterial endocarditis); mitral disruption with calcification, three cases; mitral valve thrombosis, four cases; and reoperation for insufficiency due to inadequate leaflet coaptation, two cases. Although we plan to continue using the Hancock valve, we are worried about what to use in the children and in patients with a small aortic root. The Hancock aortic valve sizes 19, 21, and 23 are stenotic. In that particular group of patients we have been using the Ionescu pericardial valve because of the much smaller residual gradient. Close to 100 such valves have been placed, with satisfactory results thus far but with a maximum follow-up of 30 months. DR. DONALD J. MAGILLIGAN Detroit. Mich.

[Slide] Our presentation of our experience to the Society of Thoracic Surgeons in January is updated by 3 months in this slide. At 6 years, 92.5% are free of failure-a figure not significantly different from the 94% free of failure reported by Dr. Oyer. We also have seen a difference in young patients, in our experience the breakoff being at 35 years or below. It is suggested in our experience, too, that the mitral valve degenerates faster than the aortic valve and in patients younger than 35 years. It has been suggested that the long-term results in Palo Alto are different from those in Detroit, but we are indebted to Dr. Oyer for showing that at 6 years they are virtually the same. The difference has been due to a misinterpretation of the leading edge of the actuarial curves. Does Dr. Oyer have any suggestions for us as to how to deal with the leading edge of this curve, since we are committed to actuarial analysis of valve data? It is easy to see large standard errors, but should we disregard the curves when the 95% confidence bands exceed 10%, 20%, 30%?

The Journal of

832

Oyer et al.

DR. HOOSHANG BOLOOKI Miami. Fla.

We are also grateful to Dr. Oyer for updating their results with xenograft valves. In Miami we have an older population, and we are concerned about using long-term anticoagulation in many instances. Therefore, from 1968 to 1972 we used homograft valves as substitutes in about 120 patients. Since 1974 we have been using xenograft valves. We recently reviewed our results and found them to be somewhat unique: We have two series of patients-one group of 123 patients with Hancock bioprostheses, operated upon from 1974 through 1978, and another group of 140 patients with Carpentier-Edwards valves, operated upon from 1975 through 1978. The two series are parallel and the study is prospective but nonrandomized. Although there is a 1 year difference in mean follow-up within the two groups of patients, we have seen eight tissue valve failures in the Hancock group within the first 3 years after implantation as compared with none in the CarpentierEdwards valve group. All the Carpentier-Edwards valves are presently in place, although we know of one patient who has a gradient of 15 mm Hg across a 31 mm mitral valve bioprosthesis. The Hancock valve failures have been due to calcification, stiffening, and rupture of the leaflets resulting primarily in valve stenosis. I would like to ask Dr. Oyer to comment about this type of early valve failure, since the mean interval from implantation to valve failure was approximately 32 months. It seems to us that these valves failed earlier than anticipated. These patients were not children, they did not have abnormal calcium metabolism, and they were not undergoing renal dialysis. We wonder if the stent design in Hancock valves or the preparation and preservation procedure has something to do with this early failure. DR. ALAIN CARPENTIER Paris. France

As Dr. Oyer pointed out, long-term durability is a major limitation associated with valvular bioprostheses. The expected reoperation rate of 30% at 10 years is compensated by a higher survival rate and a better quality of life associated with this technique. However, the higher calcification rate in children is of major concern. We have been studying this problem in our research laboratory during the past 3 years, and I would like to mention some preliminary results which may have practical significance. We submitted three groups of rats to high calcium, normal calcium, and low calcium diets, 10,6, and 2 gm/day, respectively. Glutaraldehyde-preserved valve tissue was subsequently implanted in the animals. We found a correlation between calcification of the implanted valves and calcium intake. Thus placing the patients with glutaraldehyde-preserved bioprostheses on a low calcium diet compatible with normal growth may prove to be a valuable approach to the problem of calcification in the future, but this theory has to be supported by further studies.

Thoracic and Cardiovascular Surgery

Incidentally, it may be of interest that the first two patients in whom we implanted a glutaraldehyde-preserved bioprosthesis are still living and well with their original valves 12 years after the operation. DR. JORDAN D. HALLER Northridge. Calif.

This seems to be an appropriate time to present follow-up studies on the low-profile bioprosthesis developed by Dr. Domingo Liotta and his associates in Buenos Aires, Argentina. This valve is mounted differently from other porcine valves, in what is believed to be a more anatomic fashion, to lessen stress and thereby improve long-term durability. Since 1976, a total of 341 valves have been implanted. Primary tissue failure occurred in four instances, an incidence in 4 years of 1.28% or 0.32% annually. Systemic embolism occurred in nine patients, an incidence of 2.88% or 0.72% annually. Patients with aortic valves are not given anticoagulant therapy. Following mitral valve replacement anticoagulation is discontinued after 3 to 6 months. Endocarditis occurred in four patients, an incidence of 1.28% or 0.32% annually. Hemodynamic studies have been done after mitral valve replacement in 14 patients, eight early postoperatively (10 to 25 days, mean 16.5 days) and six late postoperatively (12 to 46 months, mean 26.5 months). The early mean pressure gradient was 2.12 ± 1.8 mm Hg at rest and 3.66 ± 2.6 mm Hg with exercise, compared to the late mean gradient of 2.50 ± 2.7 mm Hg at rest and 5.25 ± 3.5 mm Hg with exercise. Early cardiac output was 5.12 ± 1.3 Llmin at rest and 6.74 ± 1.86 Llmin with exercise, compared to the late cardiac output of 5.11 ± 1.31 Llmin at rest and 7.57 ± 2.0 Llmin with exercise. DR. 0 Y E R (Closing) I appreciate the comments. First, I would answer Dr. Kennedy by saying that we have not looked closely in a systematic way at the ultrastructure of all valves removed. As he suggests, however, it is reasonable to assume that the various components of the valve tissue, including elastin, collagen, and mucopolysaccharides, may well be altered during the fixation process so as to affect adversely the long-term durability of these valves. As a corollary, changes in fixation processes conceivably could further improve the observed durability. It is unfortunately one of those questions that is not amenable to answer in accelerated testing models and thus represents a difficult and time-consuming area of investigation. In regard to Dr. Duran's question about the small aortic root, our feeling is that the number of patients who truly have a small aortic root in comparison to their body size is fairly small. The patients in whom the small aortic root is a problem are the younger patients who hope later to lead physically active lives. Such young patients with relatively long life expectancies probably stand to gain the most from the use of a bioprosthetic valve, by virture of its low thromboembolic rate and lack of need for anticoagulation. Our recommendation in

Volume 80 Number 6 December, 1980

these patients is to perform an anulus-enlarging procedure and implant a bioprosthesis. Inasmuch as older patients generally lead more sedentary life-styles, not often requiring cardiac outputs substantially increased over resting levels, anulus enlargement is usually unnecessary. In these patients we use the small bioprosthesis, with the realization that relatively large gradients may occur during transient periods of exercise, but that during usual daily activities only minor gradients will exist. Dr. Magilligan inquired about the reporting of standard data which have large associated errors. We have elected not to report actuarial data points with standard errors of the mean greater than ±5%. The reason for that policy is that the statistical uncertainty regarding the true value of such data points renders them, at best, only marginally useful or possibly misleading for drawing conclusions on which clinical decisions or therapy is based. I cannot explain Dr. Bolooki's observed difference in

Hancock porcine bioprosthetic valve

833

valve failure between the Carpentier-Edwards and the Hancock valves. Closer inspection of his follow-up data in all regards would be required to determine the significance of this observation. It is a pleasure to have Dr. Alain Carpentier's comments. He has been one of the pioneers in the area of valvular bioprostheses and appears to be in the forefront again, investigating the mechanisms of failure of the xenograft bioprosthesis. It is interesting that he notices an association between dietary calcium levels and calcification of gluteraldehyde-fixed tissues. I understand that in his experimental studies he has implanted fixed valve-leaflet segments in the subcutaneous tissues; I am not sure that that model system allows a fair comparison with tissue that is in a rapidly flowing bloodstream, particularly insofar as the effects of cellular elements attaching to and migrating into the implanted foreign tissue may differ in the two locations. Nevertheless, this remains an area clearly in need of further investigation.