Transmission of Hepatitis A to Patients with Hemophilia by Factor VIII Concentrates Treated with Organic Solvent and Detergent To Inactivate Viruses

Description of the Outbreak

The first 3 cases of hepatitis A were diagnosed in 1989, 10 cases were identified in 1990, 33 in 1991, and 6 in 1992 at 12 hemophilia centers located in Milan, Trent, Vicenza, Castelfranco Veneto, Bologna, and Pavia in Northern Italy; in Florence, Rome, and Perugia in Central Italy; and in Naples and Bari in Southern Italy. All but 1 of the 52 case patients had severe hemophilia A, and their median age was 24 years (range, 2 to 43 years). Most case patients were identified because jaundice developed (42 of 52 patients; 81%), and hepatitis A was diagnosed when serum samples were positive for IgM anti-HAV antibodies (HAVAB-M EIA; Abbott Laboratories, North Chicago, Illinois). The course of hepatitis was uncomplicated and all patients recovered completely.

Within 2 months before the onset of jaundice (the upper limit of the incubation period of hepatitis A), all patients had been infused with a factor VIII concentrate manufactured in Italy from pooled plasma collected in the United States from paid plasmapheresis donors. No blood product other than factor VIII concentrate had been administered to the index case subjects during that period. A virucidal step based on treatment with an organic solvent and a detergent (tri-[n-butyl]-phosphate/polysorbate 80) was incorporated into the manufacturing process of the concentrate [2], followed by an ion-exchange chromatography step to purify factor VIII [3]. At least 20 different concentrate lots were administered to case patients in the 2-month period before jaundice developed, 12 of them to more than 1 patient. Hepatitis A has not been reported among patients with hemophilia receiving factor VIII concentrates treated with other virucidal methods, such as pasteurization [8] or vapor heating, or among patients with hemophilia B (who, in Italy, almost exclusively receive vapor-heated concentrates). No additional cases have been reported since June 1992, but most Italian patients with hemophilia have not been given the solvent-detergent-treated concentrate, and many of them have been vaccinated against hepatitis A since late 1992, when the hepatitis A vaccine became available in Europe.

Methods

Case-Control Study

The first 29 consecutive patients with hepatitis A and jaundice were enrolled. One to three nonjaundiced control patients with hemophilia (n = 71), randomly chosen from the national registry of patients with congenital coagulation disorders (Istituto Superiore di Sanita, Rome, Italy), were matched to each case patient by age, province of residence, hemophilia center, type and severity of hemophilia, and exposure to factor VIII concentrate in the 2 months before the onset of jaundice. For both case patients and control patients we used a questionnaire to gather the following information about the 2-month period before the onset of jaundice in case patients: lots of concentrate and the corresponding virucidal methods used to treat them, number of doses infused, contact with persons who had jaundice or hepatitis A, travel to countries reported to have a high attack rate for hepatitis A, and consumption of raw shellfish. None of the control patients were vaccinated against hepatitis A at the time of the survey.

Statistical Analysis

Odds ratios and 95% CIs were calculated for matched and unmatched analyses [9].

Polymerase Chain Reaction Analyses

The RNA was prepared by a modification of the method of Chomczynski and Sacchi [10]. Briefly, 100 µL of serum or 100 µL of factor VIII concentrate that had been reconstituted with sterile water to a 10-fold concentration was mixed with 400 µL denaturing buffer (5.2 mol/L guanidinium thiocyanate, 0.5% N-lauryl sarcosine, and 0.025 mol/L TRIS, pH 8.0), 50 µL (10-fold concentration) phenol extraction buffer (1 mol/L TRIS, pH 8.0; 0.1 mol/L EDTA; and 10% sodium dodecyl sulfate), and 10 µg glycogen (Boehringer Mannheim Corp., Indianapolis, Indiana), extracted twice with phenol-chloroform (at 65 °C for 30 minutes; at room temperature for 5 minutes) and then with chloroform. After precipitation with isopropanol, RNA pellets were resuspended in 10 µL water and the entire sample was used for the synthesis of cDNA.

Nested primer sets were used [11]. Set one included 5′-AGTGCAGTCAACTTTGAG (positions 1961 to 1978) and 5′-ATCTGGAACATTCTGTTCTG (positions 2267 to 2248), and 5′-ACAGGTATACAAAGTCAG (positions 2020 to 2037) and 5′-CTCCAGAATCATCTCC (positions 2226 to 2211). Set two included 5′-TCCCAGAGCTCCATTGAA (positions 2984 to 3001) and 5′-CATTATTTCATGCTCCTCAG (positions 3284 to 3265), and 5′-CAAATGCCATGTTATCCACTG (positions 3004 to 3024) and 5′-GGTGGAAGTGCTTCATTTGAC (positions 3211 to 3191).

Before cDNA synthesis, each 10-µL RNA sample was annealed with 10 pmol of the reverse external primer at 65 °C for 3 minutes. Synthesis of cDNA was performed in a 20-µL reaction containing the annealed primer-template, 40 U RNasin (Promega Corp., Madison, Wisconsin), 8 U avian myeloblastosis virus reverse transcriptase (Promega), 1 mmol each of four deoxynucleoside triphosphates (Pharmacia, Piscataway, New Jersey), and polymerase chain reaction buffer (Perkin-Elmer Cetus, Norwalk, Connecticut) at 43 °C for 60 minutes. The polymerase chain reaction was performed with each cDNA sample in a 100-µL reaction containing 50 pmol of each primer, 2.5 U Taq polymerase (Amplitaq, Perkin-Elmer Cetus), 0.2 mmol each of four deoxynucleoside triphosphates and polymerase chain reaction buffer, and overlaid with mineral oil. The first amplification was for 36 cycles at 94 °C for 1 minute, at 40 °C for 30 seconds, and at 72 °C for 1 minute. Ten µL was then amplified with the inner set of primers for 45 cycles at 94 °C for 1 minute, at 37 °C for 30 seconds, and at 72 °C for 1 minute. DNA products were analyzed by electrophoresis through 2% agarose gel and ethidium bromide staining, followed by photography under ultraviolet light.

During RNA extraction and polymerase chain reaction amplification, special precautions were taken to avoid false-positive results [12]. In every experiment, one negative control sample was processed and tested in parallel with each test sample. Every sample was tested at least twice, and results were considered valid only if confirmed in repeated experiments.

To determine the sensitivity of our polymerase chain reaction assay, we analyzed 10-fold serial dilutions of a laboratory stock of a cell culture-adapted strain of HAV [13]. This HAV stock, harvested from virus-infected cells in culture, contained 10⁷ 50% tissue culture infective doses (TCID₅₀) per milliliter of virus, as determined from immunofluorescence assays (data not shown). Hepatitis A viral sequences were detected consistently in 100-µL samples of stock diluted to 10⁻⁸, and therefore we estimated that the sensitivity of our assay was 0.01 TCID₅₀.

Determination of Nucleotide Sequences

Polymerase chain reaction products were purified by fractionation through low-melting-temperature agarose. The DNA bands were excised and DNA was extracted [14]. Sequencing was done with Sequenase (United States Biochemical, Cleveland, Ohio) as previously described [15].

Animal Studies

Seronegative juvenile chimpanzees were infused with selected lots of factor VIII, and blood samples were drawn weekly and tested for serum levels of alanine aminotransferase, isocitrate dehydrogenase, and γ-glutamyltransferase by standard methods (Metpath, Rockville, Maryland), for antibodies to HAV and hepatitis C virus by commercial assays (Abbott), and for antibodies to hepatitis E virus by an enzyme-linked immunosorbent assay using expressed hepatitis E virus nucleocapsid protein [16].

Results

Case-Control Study: Implications of Factor VIII from a Specific Manufacturer

Case patients ranged in age from 10 to 42 years (median, 22 years), control patients from 11 to 49 years (median, 22 years). Nine of 29 case patients and 19 of 71 control patients were seropositive for antibody to human immunodeficiency virus.

Case patients were more likely to have received the Italian solvent-detergent-treated factor VIII concentrate than were control patients (odds ratio, ∞; CI, 4.5 to ∞) and to have been given larger doses of concentrate (Table 1). On the other hand, case patients were neither more nor less likely than control patients to have traveled to countries known to have high hepatitis A attack rates, to have consumed raw shellfish, or to have had contact with persons with jaundice (Table 1).

Detection of Identical HAV Genomic Sequences in Patients' Serum Samples and Corresponding Factor VIII Lots

Twelve lots of factor VIII concentrate were tested using the polymerase chain reaction for HAV genomic sequences. Nine of these lots had been implicated in the transmission of HAV to 4 patients Table 2, Table 3. The remaining 3 lots, 0553901004, 0553890902, and 441001, were randomly chosen from a batch of 38 lots administered to 30 patients during a prospective safety study in Italy, in which IgM anti-HAV and hepatitis developed in 1 patient (Table 3). Serial serum samples from patients 1 and 4 (1 from Milan in Northern Italy and 1 from Bari in Southern Italy) were available, including samples collected before and after seroconversion and clinical evidence of hepatitis. These were also tested using the polymerase chain reaction (Table 4).

Factor VIII preparations were screened for the presence of HAV using polymerase chain reaction amplification of a conserved region of the structural gene, VP3. Hepatitis A viral sequences were detected in 5 of the 12 lots of factor VIII concentrate tested (Table 3). Specifically, of the 9 lots implicated in the 4 cases of hepatitis A, viral sequences were detected in the only lot administered to patient 1, in 1 of 2 lots administered to patient 2, in 1 of 2 lots given to patient 3, and in 1 of 4 lots given to patient 4. The remaining positive factor VIII sample was from lot 441001, 1 of the 3 lots randomly chosen from the factor VIII safety study.

Nucleotide sequence analysis of the amplified VP3 regions showed three distinct but closely related sequences from the 5 lots of factor VIII (Figure 1). Nucleotide sequences amplified from lots 0559910102 and 0556900101 were identical, as were sequences from lots 451009 and 441001. The sequence from lot 461059 was unique. The three distinct sequences identified shared 96% to 99% homology. To confirm the specificity of our polymerase chain reaction assay and to preclude the possibility of cross-contamination of samples with HAV from the laboratory, the amplified HAV sequences were compared with sequences of two strains studied in the laboratory (at the National Institutes of Health, Bethesda, Maryland) in which the polymerase chain reaction tests were done. The VP3 sequences isolated from the 5 factor VIII concentrates differed from HM-175 [17], our laboratory strain of human HAV, by 7, 8, or 9 nucleotides (88.5% to 91% homology). Similarly, the sequences differed from our laboratory strain of simian HAV, AGM-27 [18], by 12 or 10 nucleotides (85% to 87% homology). Despite the nucleotide differences detected in this region of the genome, the amino acid sequences were completely conserved among all 5 strains compared in Figure 1. The VP3 sequences were also compared with those of 16 other strains from 9 countries, including the United States (8 strains) and Italy (1 strain). The VP3 sequence identified in factor VIII lots 0559910102 and 0556900101 was unique, as was that from factor VIII lot 461059. Of the 16 strains screened, the closest match differed from the sequence in factor VIII lots 0559910102 and 0556900101 by 2 of 79 nucleotides and from that of factor VIII lot 461059 by 1 of 79 nucleotides. The sequence obtained from factor VIII lots 451009 and 441001 was more common and was identical to that identified in strains CARI (Italy), NC-1 (North Carolina), MS-1 (New York), and KMW-1 (Switzerland) [11].

To obtain molecular evidence that the HAV detected in the factor VIII preparations was responsible for transmission of HAV to patients 1 and 4, serial serum samples from both patients were tested in the polymerase chain reaction assay. In both patients, HAV sequences were detected in serial serum samples spanning a period of 6 to 7 weeks (Table 4). Nucleotide sequence analyses of all polymerase chain reaction products from the serum samples of patient 1 consistently showed VP3 sequences identical to the sequences detected in factor VIII lot 0556900101. Similarly, nucleotide sequence analysis of polymerase chain reaction products from the serum samples of patient 4 showed VP3 sequences identical to those detected in factor VIII lot 461059.

To obtain additional evidence that the HAV genomes detected in the factor VIII lots were indeed identical to those detected in the corresponding patients' serum samples, we tried to amplify a second region of the HAV genome, VP1/2A. Successful amplification of the less conserved VP1/2A region was achieved only for factor VIII lot 0556900101 and the serum sample from patient 1 (Table 4). Nucleotide sequence analysis of a 46-base-pair stretch of this more variable region again revealed identical HAV sequences from serum samples of patient 1 and factor VIII lot 0556900101 (Figure 1).

Inability To Induce Hepatitis A in Chimpanzees by Infusion of Factor VIII

Selected lots of factor VIII concentrates were administered intravenously to three chimpanzees to recover infectious HAV. The first chimpanzee received the equivalent of one-half ampule (10 mL = 250 U) of lot 0556900101, which had been given to patient 1, and the second chimpanzee received the equivalent of one-half ampule each (5 mL = 500 U each) of lots 0559910101 and 0559910102, which had been given to patient 2. The third chimpanzee received the equivalent of 20 ampules (200 mL = 20 000 U) of lot 461059, which had been given to patient 4. The chimpanzees were observed for as long as 10 months.

Evidence of HAV infection did not develop in the three chimpanzees, and all remained seronegative for anti-HAV, although elevated liver enzyme values developed during weeks 4 and 5 after inoculation in the chimpanzee that received lot 0556900101. These elevated values were accompanied by gastrointestinal bleeding of unknown cause. The animal did not seroconvert for antibody to any of the recognized hepatitis viruses, including hepatitis E. Results of liver biopsy samples obtained at this time appeared normal, and the enzyme abnormalities were believed to be not of infectious, and perhaps not of hepatic, origin (data not shown).

Discussion

Identification of HAV viral sequences in 5 of 12 lots of factor VIII tested indicates that viral contamination of factor VIII prepared by the solvent-detergent-chromatographic procedure occurred relatively frequently. Epidemiologic data, coupled with sequence analysis of viral genomes amplified both from factor VIII lots and from the serum samples of patients receiving those lots, provided convincing circumstantial evidence that the hepatitis was due to virus acquired by infusion of factor VIII. Although the noted differences in sequence were small, they provided reliable identification markers because the sequence data were obtained by direct sequencing of polymerase chain reaction product and, hence, represented the consensus sequence for each sample. The discovery of a VP3 sequence unique to patient 4 and one of the factor VIII lots with which he was infused suggests that he could have acquired HAV from infusion of that lot of factor VIII. Even more compelling evidence was obtained for patient 1, who received a single lot of factor VIII: Identical viral sequences in both the conserved VP3 and more variable VP1/2A regions were found in the factor VIII and the patient's serum samples.

Unfortunately, we were unable to infect three chimpanzees with implicated lots of factor VIII. The inability to show transmission of HAV in the first two chimpanzees (infused with one-half ampule volumes) was probably due to the small amount of inoculum available for infusion. The third chimpanzee was infused with 20 000 U of factor VIII lot 461059 (patient 4 received 8000 units of this lot). The failure to show infectivity, in this case, suggests that the number of infectious virus particles was so low that only random samples of the lot contained viable virus. However, epidemiologic evidence strongly suggests that enough viable virus was present in the factor VIII to infect humans and that its level of infectivity for patients with hemophilia was consistent with the chimpanzee data (<1 infectious dose/20 000 units of clotting factor).

The case–control study indicated that it was unlikely that person-to-person transmission or contamination of food or water caused the outbreak of hepatitis A in Italian patients with hemophilia. We found a strong association between the occurrence of hepatitis and use of a high-purity factor VIII concentrate manufactured in Italy, which was similar to concentrates produced in Austria and Germany and administered to the patients with hemophilia in Germany, Ireland, and Belgium in whom hepatitis A developed. All concentrates were produced by ion-exchange chromatography, contained little immunoglobulin or plasma proteins other than factor VIII and von Willebrand factor [3], and were virally inactivated by the addition of a solvent-detergent mixture [2]. The concentrates differed, however, in that the Italian product was manufactured from plasma obtained in the United States from paid plasmapheresis donors [1], whereas the German, Irish, and Belgian products were manufactured from plasma of local unpaid donors [5-7]. The epidemiologic association of the factor VIII lots with transmission of HAV, coupled with the detection of identical HAV sequences in the factor VIII and the corresponding patient's serum samples in two cases, strongly supports the conclusion that HAV was acquired from the factor VIII preparations.

Although the epidemiologic circumstances indicate that the Italian outbreak of hepatitis A was parenterally acquired, it has not been determined whether the plasma used to manufacture the concentrate was originally contaminated with HAV or whether contamination occurred during fractionation. Inspection of the manufacturer's facilities and water supply by experts of the Italian Ministry of Health failed to identify any possible sources of contamination, and cases of hepatitis A were also associated with concentrates manufactured in Austrian and German plants [5-7]. Twenty lots manufactured during a period of 2 years were associated with the cases of hepatitis in Italy, making occasional contamination of the plant or a single error during manufacturing of the concentrate unlikely.

Epidemiologic considerations are consistent with the possibility that HAV-infected plasma units may have contaminated the pools used to prepare the concentrates. The attack rate of HAV infection is decreasing in all Western countries [19, 20], and more susceptible patients with hemophilia now comprise the population of clotting factor recipients. Also, persons susceptible to HAV infection are more likely to reach the age at which they might donate plasma during the asymptomatic period of viremia associated with HAV infection. We have shown previously that human viremic sera or plasma may contain 10³ to¹⁰⁵ chimpanzee infectious doses of HAV per milliliter [21]. If a contaminated plasmapheresis unit were to be included in a lot of pooled plasma for fractionation, the resultant titer of virus could be 0.1 to 10 infectious doses per milliliter of plasma. If viremic donations contaminate a plasma pool, the content of specific neutralizing antibodies, which is decreasing because of the changing epidemiologic features of HAV infection, may not be sufficient to neutralize HAV. Alternatively, virus-antibody complexes might dissociate during manufacturing of the concentrate, perhaps due to the chromatographic steps that remove immunoglobulins and other proteins from the final product or to the dissociating effect of solvent and detergent.

In this study, three different strains of HAV were recovered from lots of factor VIII. This suggests that they were not contaminated from the same plasmapheresis donor, but, in fact, the strains were closely related and could have represented variants of HAV within the same population of HAV virions or, alternatively, viruses from plasmapheresis units obtained from different donors whose exposure was epidemiologically related. Further information on the donors is not available. However, comparison of these strains with other strains studied previously provides convincing evidence that the viruses recovered from factor VIII concentrates and the patients with hemophilia were not contaminants within the laboratory but unique strains that are not widely distributed. Finally, the duration of the viremia in the two HAV-positive patients with hemophilia was surprising but perhaps not completely unexpected because viremia has been detected for as long as 28 days before onset of hepatitis in enterically infected persons [22]. It is important to note that this is the first determination of HAV viremia using the polymerase chain reaction. The polymerase chain reaction was shown to be more sensitive than infectivity for detecting hepatitis C virus viremia after onset of the immune response, possibly because of neutralization of virus by antibody [23]. It must be determined if this is the case in HAV infections.

In conclusion, patients with hemophilia who are treated with large-pool concentrates are at risk for infection with HAV. Thus, susceptible patients should be vaccinated against HAV before they are given solvent-detergent-treated concentrates [24]. However, other blood-borne, nonenveloped viruses, such as parvovirus B19 [25] and hepatitis E virus, could pose similar risks to these patients. Vaccines are not yet available for these agents.