Long-Term Efficacy and Immune Responses following Immunization with the RTS,S Malaria Vaccine

Transcription

1 1139 Long-Term Efficacy and Immune Responses following Immunization with the RTS,S Malaria Vaccine J. A. Stoute, K. E. Kester, U. Krzych, B. T. Wellde, Departments of Immunology, Entomology and Membrane Biochemistry, T. Hall, K. White, G. Glenn, C. F. Ockenhouse, Division of Communicable Diseases and Immunology, Walter Reed N. Garcon, R. Schwenk, D. E. Lanar, P. Sun, P. Momin, Army Institute of Research, Washington, DC; SmithKline Beecham Biologicals, Rixensart, Belgium R. A. Wirtz, C. Golenda, M. Slaoui, G. Wortmann, C. Holland, M. Dowler, J. Cohen, and W. Ripley Ballou The malaria sporozoite vaccine candidate RTS,S, formulated with an oil-in-water emulsion plus the immunostimulants monophosphoryl lipid A and the saponin derivative QS21 (vaccine 3), recently showed superior efficacy over two other experimental formulations. Immunized volunteers were followed to determine the duration of protective immune responses. Antibody levels decreased to between one-third and one-half of peak values 6 months after the last dose of vaccine. T cell proliferation and interferon-g production in vitro were observed in response to RTS,S or hepatitis B surface antigen. Seven previously protected volunteers received sporozoite challenge, and 2 remained protected (1/1 for vaccine 1, 0/1 for vaccine 2, and 1/5 for vaccine 3). The prepatent period was 10.8 days for the control group and 13.2 days for the vaccinees (P õ.01). Immune responses did not correlate with protection. Further optimization in vaccine composition and/or immunization schedule will be required to induce longer-lasting protective immunity. We recently reported the results of a preliminary clinical (HBsAg) epitopes during the interval between the two challenges. trial of the Plasmodium falciparum circumsporozoite protein We report here the results of these studies. (CS) based vaccine RTS,S, in which 3 groups of malarianaive volunteers were immunized with three different vaccine formulations containing increasingly more potent and complex Methods adjuvants [1]. The unprecedented level of protection (6 of 7 Vaccines. The RTS,S vaccine consists of two polypeptide volunteers protected) achieved with one of these formulations chains, RTS and S. The RTS polypeptide contains aa of (vaccine 3) was a significant step towards the goal of devel- the P. falciparum CS of the clone 3D7 fused to the surface antigen oping an effective malaria vaccine and gave us an opportunity (S) of the hepatitis B virus (adw serotype). The CS portion of the to assess the duration of vaccine-induced immunity. To be fusion protein contains the final 19 NANP tetrapeptide repeats and useful, such a vaccine should confer protection for a reasonable all of the carboxy-terminal sequence up to the transmembrane period. Six months between reexposure to sporozoites was region, including TH2R, TH3R, and the putative universal T cell thought to represent a clinically relevant interval. We therefore epitope [2]. The S polypeptide portion of the fusion protein consists of 226 aa of HBsAg. The RTS fusion protein and the S polypeptide asked all of the previously protected volunteers to undergo a are coexpressed in Saccharomyces cerevisiae and spontaneously second homologous challenge by exposure to P. falciparum assemble into virus-like particles referred to as RTS,S. These puriinfected mosquitoes. We also measured a number of immune fied particles constitute the antigen used in these formulations. responses to CS protein and hepatitis B surface antigen Three different formulations were studied (vaccines 1 3). Vaccine 1 contained 50 mg of RTS,S per 1-mL dose and was formulated with SBAS4 (aluminum hydroxide and 3-deacyl-monophosphoryl lipid A [MPL]). Vaccine 2 contained 50 mg of RTS,S Received 29 September 1997; revised 24 March per 0.5-mL dose in SBAS3 (an oil-in-water emulsion). Vaccine 3 The views of the authors do not purport to reflect the position of the Departcontained 50 mg of RTS,S per 0.5-mL dose in SBAS2 (the SBAS3 ment of the Army or the Department of Defense. The US government has the right to retain a nonexclusive, royalty-free license in and to any copyright emulsion plus MPL and the saponin derivative QS21 [3]). covering this paper. Subjects and study design. The first portion of the trial includ- Volunteers were recruited according to US Army regulations (AR 70-25, ing the first challenge was designed as an open-label study without Use of Volunteers as Subjects of Research) under a protocol approved by the placebo control, the details of which have been published else- Walter Reed Army Institute Human Use Review Committee and the Army Surgeon General s Human Subjects Review and Regulatory Board. Written where [1]. Briefly, 46 malaria-naive volunteers aged years informed consent was obtained from all volunteers before study entry. were recruited by noncoercive means. The subjects were assigned Financial support: US Army Medical Research and Materiel Command. to receive one of the three vaccines. Three doses of each vaccine Reprints or correspondence (present address): Dr. José A. Stoute, US Army were given in a schedule of 0, 4, and weeks. All vaccines Medical Research Unit, Nairobi, Kenya, Unit 64109, Box 401, APO AE were given by intramuscular injections, with each dose adminis (stoutej@wrsmtp-ccmail.army.mil). tered in alternate deltoid muscles. In an attempt to reduce side The Journal of Infectious Diseases 1998;178: This article is in the public domain. effects observed after the second dose in 2 volunteers, the third doses of vaccines 2 and 3 were decreased to 0.1 ml [1]. Forty-

2 1140 Stoute et al. JID 1998;178 (October) one volunteers received a second dose of vaccine, and 27 received Ninety-six-hour supernatants from parallel cultures of selected a third dose. The first sporozoite challenge took place 3 weeks samples were collected to measure interferon (IFN)-g by ELISA after the third dose of vaccine. (Genzyme, Cambridge, MA). Serial 2-fold dilutions of a human The second portion of the trial involved the rechallenge and IFN-g standard were assayed in parallel, and concentrations were took place Ç6 months after the initial challenge (week 54). Six calculated from the standard curve. unimmunized malaria-naive control volunteers were used for each Phenotypic analysis of T cells responsive to RTS,S was done challenge to establish the infectivity of the sporozoites. Four weeks on samples obtained before rechallenge by cytofluorometry using before the second challenge, the control volunteers received 0.5 labeled monoclonal antibodies against CD4, CD8, and CD16. ml of a placebo formulation containing the adjuvant SBAS2 withex PBMC obtained from the 6 protected volunteers were characterized out RTS,S antigen by intramuscular injection. One of these 6 vivo and PBMC from 5 of these 6 were cultured for 8 9 days volunteers declined to participate in the challenge and was replaced in the presence of 5 10 mg/ml RTS,S to determine the ratio of by a malaria-naive volunteer who did not receive the placebo. CD4, CD8, and CD16 lymphocytes in blast or resting cell popula- Follow-up. Blood for measurements of antibodies and T cell tions. Blast and resting populations were identified by forward and responses was obtained at baseline and after each immunization sideward scatter. as previously reported [1]. Beginning 2 months after the first sporoexpanded Sporozoite challenge. Cloned P. falciparum 3D7 parasites were zoite challenge, all volunteers who had received three doses of from a master seed lot and used to infect laboratory-reared vaccine, including the 9 volunteers who were protected at the first Anopheles stephensii mosquitoes. Mosquitoes harvested from the challenge, were followed to obtain blood for additional antibody same batch fed on each challenged subject for 5 min. Those mosqui- and T cell studies. toes with blood meals were dissected to quantify the presence of Measurement of antibody responses. Blood was allowed to viable sporozoites. Challenge continued until 5 infected mosquitoes clot, and sera were separated and stored at 070 C until needed. had successfully fed. The volunteers were challenged over a period Total IgG against the NANP repeat was measured by ELISA using of 2 consecutive days, and the vaccinees and control volunteers the recombinant molecule R32LR that contains the sequence were equally allocated for challenge between the 2 days. Volunteers [NVDP(NANP) 15 ] 2. Anti-repeat IgG levels were determined relasigns and symptoms of malaria and to obtain thick and thin Giemsa were observed daily from days 7 to 21 after challenge to detect tive to a CS-specific standard [4]. Anti-repeat IgM levels were determined by a similar protocol with a 1:2000 dilution of horseexamined for 200 high-power fields before declaring them negative, smears. Thick smears from asymptomatic subjects were routinely radish peroxidase labeled anti-human IgM (Kirkegaard & Perry, Gaithersburg, MD) as the secondary antibody. A human monoclinically. After day 21, routine daily smears were discontinued for but smears were reviewed exhaustively if malaria was suspected clonal IgM antibody that recognizes the CS repeat sequence was used as a positive control for the IgM ELISA. To measure antibodasymptomatic volunteers who had not developed a patent infection. These volunteers were followed clinically on a daily basis for an ies to the nonrepeat C-terminal portion of the CS protein, aa 318 additional week and then weekly for another 4 weeks. Volunteers 405 of CS were expressed as a fusion protein (PfCS) with thiorewho remained parasite-free after this period were declared protected. doxin in the pet32 plasmid Escherichia coli expression system Treatment of unprotected volunteers with a standard regimen of (Novagen, Madison, WI). Wells were coated with 0.1 mg/ml purichloroquine was initiated within several hours after a parasitologic fied PfCS or thioredoxin as control. In all cases, reactivity to diagnosis of malaria was established. Three daily consecutive negathioredoxin was negligible, and therefore, this value was not subtive blood smears were required before declaring a subject cured of tracted from the optical density (OD) measured using the fusion malaria following treatment. protein. Antibodies to HBsAg were measured by ELISA [5]. Antirepeat IgG levels were expressed as micrograms per milliliter, and anti-repeat IgM and anti C-terminal antibody levels were expressed as OD units (i.e., the dilution that gives an OD of 1), Results based on the average value derived from three consecutive dilutions Safety and reactogenicity of formulations. The safety and over a linear range of the dilution curve. Sera were also reactogenicity data were previously reported [1]. There were analyzed by indirect fluorescent antibody assay using air-dried P. no additional doses of vaccine administered between the first falciparum sporozoites [6]. and second challenge. The placebo formulation was well-toler- Measurement of T cell responses. Peripheral blood mononu- ated, and the only reactions were mild soreness at the site of clear cells (PBMC) were isolated from donor blood by gradient injection. centrifugation on ficoll and were stored in liquid nitrogen until use Sporozoite rechallenge of previously protected volunteers. in some assays (weeks 0 6) or were used fresh for others (weeks Of 22 subjects who participated in the first challenge (8 re ). For proliferative assays, cells were washed, diluted in ceived vaccine 1, 7 received vaccine 2, and 7 received vaccine culture medium, adjusted to a concentration of /ml, and 3), a total of 9 volunteers were protected (1 from vaccine 1, 2 dispensed in 0.1 ml into triplicate wells of 96-well round-bottom plates. Purified recombinant RTS,S (10 mg/ml) and HBsAg (10 from vaccine 2, and 6 from vaccine 3). Seven of these immune mg/ml) were used to stimulate cell cultures for 7 days. Control volunteers agreed to participate in a rechallenge 6 months later. cultures were stimulated with phytohemagglutinin (2 mg/ml) or In both challenges, all of the controls developed patent para- medium alone. Proliferative responses were measured by uptake sitemia. Volunteer 2 was the only vaccine 1 recipient who was of tritiated thymidine, and results were calculated as stimulation protected in the first challenge, and this volunteer remained indices. A stimulation index 3 was considered positive. protected on rechallenge. In contrast, the volunteer who re-

3 JID 1998;178 (October) RTS,S Malaria Vaccine 1141 consistently stronger than were responses to HBsAg (not shown). Proliferative responses fluctuated during the 6-month interval between the two challenges but clearly remained positive at the time of rechallenge. Although there was a suggestion that the first sporozoite challenge might have boosted CS responses in some subjects, the data are too limited to be considered conclusive. Similarly, the detection of IFN-g in cell culture supernatants obtained just before both challenges did not correlate with protection. Subset analysis of these cells revealed that those undergoing blastogenesis following in vitro stimulation with RTS,S were predominantly CD4 T cells (table 3). Discussion Figure 1. Malaria-free survival in controls and vaccinees after secthe The RTS,S malaria vaccine formulations we studied were designed to induce strong immune responses to the P. falciparum CS antigen. Our data indicate that antibody and cellular responses elicited by RTS,S were largely sustained throughout ond challenge. 6 months of follow-up reported here. Antibody levels against CS repeat epitopes did decrease from the peak levels achieved just before the initial sporozoite challenge, but they ceived vaccine 2 and 4 of 5 volunteers who received vaccine still remained at relatively high levels at the time of rechallenge. 3 developed malaria on rechallenge. Considering all volunteers There was no clear evidence of boosting of CS antibodies by who became infected, the mean prepatent period was 10.8 days the first challenge, but we cannot exclude this possibility on for the control group and 13.2 days for the vaccinees, and this the basis of the limited sampling done in the period just after increase was statistically significant (P õ.01, log-rank test). this challenge. The kinetics of the cellular responses were more The Kaplan-Meier survival curve for this challenge is shown suggestive of boosting, but in the absence of adequate data in figure 1. from nonimmunized sporozoite-challenged controls, this also Kinetics of antibody responses. Figure 2 summarizes the remains speculative. anti-repeat IgG and anti hepatitis B antibody responses over In the face of reasonably strong immune responses, we were time as measured by ELISA. Over a period of 24 weeks disappointed in the relative decrease in vaccine efficacy between following the third dose of vaccine, geometric mean antirepeat the two challenges. However, we cannot exclude the IgG levels decreased from 7 to 2.8 mg/ml for vaccine possibility that the outcome reflected unrecognized but significant 1 recipients and from Ç50 to 30 mg/ml for vaccines 2 and differences between the two challenges. Despite extensive 3 recipients. Geometric mean anti-hbsag antibody levels de- procedures designed to standardize the challenge model, we creased from 18,000 to 5000 miu/ml for vaccine 1 recipients can only be sure that a minimally infectious challenge has and from 29,000 to 10,000 miu/ml for vaccines 2 and 3 been done when all control volunteers become infected. It is recipients. No significant boosting of anti-repeat antibody responses impossible to exclude an inoculum effect caused by unrecog- was observed after challenge. However, any boosting nized differences in the batch of infected mosquitoes used for effect could have been missed in the 2-month interval between the second challenge. If these effects resulted in either increased the first challenge and subsequent antibody determinations. infectivity or greater numbers of sporozoites injected during The apparent rise in anti-repeat and anti-hbsag antibody the second challenge, the apparent loss of efficacy would be levels for vaccine 3 recipients at week 42 and in anti-hbsag compounded by the lower antibody and T cell responses present antibody levels for vaccine 2 recipients at week 54 (figure 2) at that time. Since it is believed that the prepatent period is is due to the missing value of a volunteer with low-level inversely related to the size of the sporozoite inoculum that antibody. Antibody responses measured by ELISA to the C- successfully completes liver-stage schizogony [7], the signifi- terminal region of CS or measured by immunofluorescent cantly prolonged malaria-free survival for the vaccinees as a antibody assay against native CS protein showed similar decreases group compared with controls indicated that the vaccine was over time (table 1). indeed partially effective, even in those who were not protected. Kinetics of T cell responses. Table 2 summarizes T cell Despite measurement of a large number of parameters, the responses to RTS,S antigen in the 7 volunteers who underwent data revealed no correlation between outcome and immune rechallenge. Proliferative responses, defined by stimulation indices responses in either challenge. As shown in tables 1 and 2, 3, were detected in all 7 rechallenged subjects at least volunteer 2 had the lowest anti-repeat antibody levels, as well once during the length of the study. Responses to RTS,S were as low IFN-g responses, at the time of the second challenge

4 1142 Stoute et al. JID 1998;178 (October) Figure 2. Geometric means of anti-circumsporozoite repeat and anti hepatitis B surface antigen (Hep B S Ag) antibody levels over time. Arrows indicate immunization. *Challenge. but nevertheless remained protected. In contrast, volunteer 25, than one potentially protective mechanism. For example, antibodies who had much higher antibody levels and greater proliferative may have been protective in one subject, whereas anwho responses and IFN-g production, was not protected, while vol- other may have been protected on the basis of cell-mediated unteer 38 had the highest antibody level and high IFN-g and killing of parasites in the liver. Indeed, RTS,S was designed was protected both times. The lack of correlation between anti- to activate both arms of the immune system, and the inclusion body responses and protection seen in this and other studies of C-terminal CS sequences containing T helper and cytotoxic [1, 4] suggests that immunization with RTS,S may induce more T lymphocyte epitopes was principally to target the intracellular

5 JID 1998;178 (October) RTS,S Malaria Vaccine 1143 Table 1. Summary of antibody responses of malaria vaccine recipients before each challenge. First challenge Second challenge Sporozoite Sporozoite Prepatent Anti-repeat C-terminal Anti-repeat IFA Prepatent Anti-repeat C-terminal Anti-repeat IFA Volunteer Vaccine period IgG (mg/ antibody IgM (OD (inverse period IgG (mg/ antibody IgM (OD (inverse ID no. group (days) ml) (OD units) units) titer) (days) ml) (OD units) units) titer) 2 1 ú , ú ú , , , , ú , , , ú ú , ,400 ú , , ú , ú , NOTE. ND Å not done; prepatent period ú60 days Å protection; OD units Å dilution that gives optical density at 414 nm Å 1; IFA Å immunofluorescent antibody assay. liver-stage parasites [8 10]. Although preclinical studies in rodents indicated that strong CS-specific CD8 cytolytic responses were achieved with the SBAS2 formulation, cytotoxic T lymphocyte activity has not been demonstrated in peripheral T cell populations from RTS,S-immunized subjects (data not shown). Further analysis of CD4 and CD8 T cell responses continues to be a major objective of ongoing studies with this vaccine. The short-term efficacy of vaccine 3 was much greater than that of the other two formulations, though all three contained the same amount of RTS,S antigen. Clearly, the choice of adjuvants plays a critical role in promoting protective responses. The link between the antigen, adjuvants, and efficacy raises a number of intriguing possibilities. It is likely that RTS,S antigen enters the circulation following intramuscular injection and finds its way to the reticuloendothelial organs, including the spleen and liver. The RTS,S antigen may actually target the liver by virtue of the presence of a hepatocyte-binding site on the S antigen [11]. If the particles travel in association with the adjuvant, potent immunostimulants known to elicit CD4 and CD8 responses, it is theoretically possible that the RTS,S particle could elicit a local microinflammatory focus in the liver in addition to the peripheral immune responses that we can measure in vitro. Such inflammation could be critical for driving cell-mediated responses capable of killing developing liver-stage parasites. The fact that the control volunteers immunized with the placebo were not protected indicates that the antigen is a vital component of this vaccine and that the adjuvant alone does not induce nonspecific immunity. Why did the efficacy of RTS,S decline in the interval between the two challenges? One possibility is that the local effects on the liver are transient and decline as the antigen is degraded and cleared. It is also possible that the decrease in the amount of third doses of vaccines 2 and 3 may have had an unanticipated negative effect on the duration of protection, since, by contrast, the vaccine 1 recipient who remained protected in the two challenges received a full third dose. Further studies are underway to investigate the impact of dose on protection. There may be several causes for vaccine failure at the level of liver-specific responses. A preponderance of data suggest that CD4 and CD8 T cells are the important effectors against liver-stage schizonts [10]. Others have suggested that Table 2. Proliferative and interferon (IFN)-g responses to RTS, S in rechallenge volunteers. Stimulation index at week IFN-g (pg/ml) at week Prepatent Volunteer Protection period Vaccine ID no * * 0 29* 54* status (days) õ20 40 õ20 P NA õ20 ú1000 ú1000 NP õ õ20 NP ND õ20 õ20 ND NP õ P NA õ NP õ NP 12 NOTE. ND: not done; NP: not protected; P: protected; NA: not applicable. * Peripheral blood mononuclear cells were collected prior to challenge.

6 1144 Stoute et al. JID 1998;178 (October) Table 3. Phenotype of gated lymphocytes from rechallenge volun- Trials, led by Moshe Shmuklarsky, for the excellent support withteers at week 54 before and after in vitro stimulation with RTS, S. out which these studies would not have been possible. We also thank Carolyn Holland, Eugene Watson, and Joe Williams for CD4 CD8 CD16 excellent ELISA support. Ex vivo cells 38.9 { { Resting cells 50 { { References Blasts 80 { { Stoute JA, Slaoui M, Heppner DG, et al. A preliminary evaluation of NOTE. Data are % (mean { SD). a recombinant circumsporozoite protein vaccine against Plasmodium falciparum malaria. N Engl J Med 1997;336: Sinigaglia F, Guttinger M, Kilgus J, et al. A malaria T-cell epitope recognized in association with most mouse and human MHC class II molecules. activated T cells are cleared from the liver in the presence of Nature 1988;336: malaria antigen by immunoregulatory mechanisms unique to 3. Kensil CR. Saponins as vaccine adjuvants. Crit Rev Ther Drug Carrier Syst 1996;13:1 55. this organ [12]. However, studies involving rodents or human 4. Gordon DM, McGovern TW, Krzych U, et al. Safety, immunogenicity, volunteers immunized with irradiated sporozoites show that a and efficacy of a recombinantly produced Plasmodium falciparum cirkey to maintaining immunologic memory and activated T cell cumsporozoite protein hepatitis B surface antigen subunit vaccine. J responses in the liver is the persistence of schizont antigen in Infect Dis 1995;171: the liver in the context of discrete foci of localized inflammaa 5. Hollinger FB, Troisi CL, Pepe PE. Anti-HBs responses to vaccination with human hepatitis B vaccine made by recombinant DNA technology in tion [13 15]. Proinflammatory cytokines in the microenvironyeast. J Infect Dis 1986;153: ment may be critical to overcoming constitutive IL-10 produc- 6. Wirtz RA, Ballou WR, Schneider I, et al. Plasmodium falciparum: immution by hepatocytes, which limits activated T cell populations nogenicity of circumsporozoite protein constructs produced in Escherichia [16]. Short-term protection following exposure to a potent coli. Exp Parasitol 1987;63: RTS,S vaccine could have led to transient induction of actimalaria. 7. Glynn JR, Bradley DJ. Inoculum size, incubation period, and severity of Analysis of data from malaria therapy records. Parasitology vated T cells in the liver; however, in the absence of persistent 1995;110:7 19. antigenic stimulation (clearance of vaccine) and despite the 8. Good M, Malow WL, Lunde MN, et al. Construction of synthetic immunopresence of circulating CS-specific IFN-g producing CD4 T gen; use of new T-helper epitope on malaria circumsporozoite protein. cells, it is likely that liver-resident CD8 memory responses Science 1987;235: failed to develop and effector responses waned. 9. Good MF, Pombo D, Quakyi IA, et al. Human T-cell recognition protein of Plasmodium falciparum: immunodominant T-cell domains map to Ongoing studies involving additional volunteers confirm the the polymorphic regions of the molecule. Proc Natl Acad Sci USA protective efficacy of the RTS,S/SBAS2 vaccine in malaria- 1988;85: naive volunteers (Kester K, unpublished data); however, the 10. Schofield I, Villaquiran J, Ferreira A, Schellekens H, Nussenzweig RS, results reported here point out that further development of the Nussenzweig V. Gamma interferon-gamma, CD8 T cells, and antibodies vaccine is needed to sustain efficacy for an extended period in required for immunity to malaria sporozoites. Nature 1987;330: Cerami C, Frevert U, Sinnis P, et al. The basolateral domain of the hepatothis population. It is possible that duration of protection may cyte plasma membrane bears receptors for the circumsporozoite protein be prolonged in populations living in a region in which malaria of Plasmodium falciparum sporozoites. Cell 1992;70: is endemic, where boosting could theoretically be achieved 12. Rajan TV. Why does Plasmodium have a pre-erythrocytic cycle? Parasitol by repeated exposure to sporozoite-infected mosquitoes. Field Today 1997;13: trials of RTS,S are underway in sub-saharan Africa to investiimmunity 13. Edelman R, Hoffman SL, Davis JR, et al. Long-term persistence of sterile in a volunteer immunized with X-irradiated Plasmodium fal- gate this possibility. Further improvements are also underway ciparum sporozoites. J Infect Dis 1993;168: aimed at improving vaccine efficacy through the addition of at 14. Stoute J, Krzych U, Ballou WR. Why does Plasmodium have a preleast one other pre-erythrocytic antigen and through the identi- erythrocytic cycle? A response. Parasitol Today 1997;13: fication of immune correlates of protection. 15. Scheller LF, Azad AF. Maintenance of protective immunity against malaria by persistent hepatic parasites derived from irradiated sporozoites. Proc Natl Acad Sci USA 1995;92: Acknowledgments 16. Alfrey EJ, Most D, Wang X, et al. Interferon-gamma and interleukin-10 messenger RNA are up-regulated after orthotopic liver transplantation We are indebted to Charles N. Oster for serving as Medical in tolerant rats: evidence for cytokine-mediated immune dysregulation. Monitor for this trial and to the staff of the Department of Clinical Surgery 1995;118: