- Email: [email protected]
WHAT DO WESTERN BLOT INDETERMINATE PATTERNS FOR HUMAN IMMUNODEFICIENCY VIRUS MEAN IN EIA-NEGATIVE BLOOD DONORS?
1023
prevalence periods.
of
urinary catheterisation
between the
two
Public Health DISCUSSION
We have shown that repeated prevalence surveys are feasible in a large general hospital, and that the results can be used to measure statistical trends in hospital infection. The British National Prevalence Survey protocol was simple to administer and use of a microcomputer for data correction and descriptive analysis allowed the production of summary statistics within a few weeks. We did not assess the reproducibility or accuracy of the method, but our results over 3 years with varying survey teams suggested that both were acceptable. This conclusion is supported by several fmdings: the risk-factors predisposing to HAI were typical 10,13 (including extremes of age, severity of underlying disease, acute or intensive medical service, long hospital stay, surgical operation, and urinary and intravenous catheterisation); the sites of HAI and the organisms involved have been reported elsewhere and were also typical; 10,21-23 and the prevalence rates of HAI and CAI were in the range expected for our patient population.6,10,23 CAI prevalence rates remained constant over the study period, but HAI rates fell after the introduction of infection control policies. We presumed that some of this reduction in HAI was due to infection control. Statistical analysis of the survey data showed that there was significant variation in risk-factors of HAI in different surveys. Analysis of the rates adjusted for risk-factors confirmed that the prevalence of HAI and HAUTI had fallen significantly after the introduction of appropriate infection control policies, independent of variations in risk. The 30% reduction in HAI following improvements in infection control in this prevalence study is similar to that found in the SENIC incidence study.6 Epidemiological criteria have been established to test a cause and effect relation.24 There should be a consistent relation between the cause and effect in different study populations; the size of the effect should be reasonably large; the strength of the association should be high, more exposure should lead to more effect; the effect should follow the intervention, and a reasonable model should exist to explain the effect.24 Freeman and McGovan4 comment that these criteria have rarely been applied to studies of infection control. The study reported here fulfils all these criteria except the first, but further work will show whether these results are repeatable in other populations. If they are, this method could be used to test the effectiveness of different infection control policies, perhaps with different hospitals or different wards acting as control sites. Repeated prevalence surveys offer a practical and sensitive way of measuring hospital infection and should be used more widely. They provide data for both infected and non-infected patients and allow adjustment of infection rates for risk. Prevalence surveys can be used to assess the impact of infection control programmes on hospital acquired infection. We believe this technique is appropriate for the analysis of hospital infection, and is especially useful in hospitals with limited resources. We
are
grateful
to
the medical and
nursing staff of the
Prince of Wales
Hospital who assisted in the surveys, and especially to our infection control Alice Chow, Christiana Chan, Deborah Ho, and Rosita Kwok. We thank Corrie Ng and Amy Cheung for computer data entry, and Linda Ip for nurses
secretarial assistance.
WHAT DO WESTERN BLOT INDETERMINATE PATTERNS FOR HUMAN IMMUNODEFICIENCY VIRUS MEAN IN EIA-NEGATIVE BLOOD DONORS?
JOAN GENESCA1 BETSY W. JETT1 JAY S. EPSTEIN2
J. WAI-KUO SHIH1 INDIRA K. HEWLETT2 HARVEY J. ALTER1
Department of Transfusion Medicine, Warren G. Magnuson Clinical Center, National Institutes of Health,1 and Division of Blood and Blood Products, Food and Drug Administration,2 Bethesda, Maryland, USA To investigate the specificity of western blot indeterminate (WBi) patterns for antibodies to the human immunodeficiency virus type 1 (HIV-1), 100 enzyme immunoassay (EIA) negative donors from whom prospectively obtained recipient sera were available were tested by WB. 20 were WBi, with p24 being the predominant (70%) and generally the only band. Among recipients of WBi blood, 36% were WBi in their 6 month post-transfusion sample, but so were 42% of a control population that had received only WB-negative
Summary
G. L. FRENCH AND OTHERS: REFERENCES 1.
2.
Haley RW, Schaberg DR, Crossley KB, von Allmen SD, McGowan JE. Extra charges and prolongation of stay attributable to nosocomial infections: a prospective interhospital comparison. Am J Med 1981; 70: 51-57. Brachmann PS. Nosocomial infection control: an overview. Rev Infect Dis 1981; 3: 640-48.
3. Daschner F. Economic aspects of hospital infections. J Hosp Infect 1982; 3: 1-4. 4. Freeman J, McGowan JE. Methodologic issues in hospital epidemiology. I. Rates, case-finding and interpretation. Rev Infect Dis 1981; 3: 658-67. 5. Tager IB, Ginsberg MB, Simchen E, Miao L, Holbrook K, Faich GA. Rationale and methods for a statewide, prospective surveillance system for the identification and prevention of nosocomical infections. Rev Infect Dis 1981; 3: 683-93. 6. Haley RW, Culber DH, White JW, et al. The efficacy of infection surveillance and control programs in preventing nosocomial infections in US hospitals. Amer J
Epidemiol 1985, 121: 182-205. Infect 1980; 1: 293-97. hospitals. Hosp J J, McGowan JE. Methodologic issues in hospital epidemiology. II. Time and accuracy in estimation. Rev Infect Dis 1981; 3: 668-77. 9. Freeman J, Hutchinson GB Prevalence, incidence and duration. Amer J Epidemiol 1980; 112: 707-23. 10. Meers PD, Ayliffe GAJ, Emmerson AM, et al. Report on the national survey of infection in hospitals, 1980. J Hosp Infect 1981; 2: (supplement). 11. Rhame FS, Sudderth WD. Incidence and prevalence as used in the analysis of the occurrence of nosocomial infections. Amer Epidemiol 1981, 113: 1-11. J 12. Aber RC, Bennett JV. Surveillance of nosocomial infection In: Bennett JV, Brachman PS, eds. Hospital infections. Boston: Little, Brown, 1979: 53-61. 13. Freeman J, McGowan JE. Risk factors of hospital infection. J Infect Dis 1973; 138: 7. Casewell MW Surveillance of infection in
8. Freeman
811-19. 14. Storch GA, Rajagopolan L. Methicillin-resistant
Staphylococcus aureus bacteraemia in children. Pediatr Infect Dis 1986, 5: 59-67 15. French GL, Cheng A, Farrington M. Prevalence survey of infection in a Hong Kong hospital using a standard protocol and microcomputer data analysis. J Hosp Infect 1987; 9: 132-42. Lowbury EJL, Ayliffe CAJ, Geddes AM, Williams JD Control of Hospital Infection London: Chapman & Hall; 1975 17. Wong ES. Guideline for prevention of catheter-associated urinary tract infections. Amer J Infect Control 1983; 11: 28-33. 18 Kahn HA. An introduction to epidemiological methods New York Oxford University Press, 1983 19. Dixon WJ, ed BMDP statistical software. Berkley: University of California Press. 16.
1985.
Armitage P. Statistical methods in medical research. Oxford: Blackwell, 1971. 21. French GL, Cheng AFB Prevalence of hospital infection at the Prince of Wales Hospital. J Hong Kong Med Assoc 1989; in press. 22 Marples RR, Mackintosh CA, Meers PD. Microbiological aspects of the 1980 national prevalence survey of infections in hospitals. J Hosp Infect 1984; 5: 172-80. 23. Haley RW, Culber DH, White JW, Morgan WM, Emori TG. The nationwide nosocomial infection rate. a new need for vital statistics. Amer J Epidemiol 1985; 20.
121: 159-67. 24. Hill AB. Principles of medical statistics. 9th ed. New York Oxford 1971, p 390
University Press:
1024 TABLE I-WB PATTERNS OF THE
blood. When serial samples from recipients with a WBi pattern were tested on two occasions, only 35% of results were reproducible. No recipient of WBi blood became EIA positive, true positive for WB, positive for HIV-1 antigen, or positive for EIA reactivity against recombinant p24 or gp41. The polymerase chain reaction was negative for gag and env HIV-1 sequences in all donors and recipients. Thus WBi patterns are exceedingly common in randomly selected donors and recipients and such patterns do not correlate with the presence of HIV-1 or the transmission of HIV-1 from donor to recipient. INTRODUCTION
WHEN the western blot (WB) assay was licensed for use as for human immunodeficiency virus type-1 (HIV-1), the criteria laid down for a positive result in the absence of risk factors or clinical suspicion were the presence of protein bands at the gag, env, and pol positions. Any pattern of bands which lacks this full array is considered WB indeterminate (WBi), although some experts regard WB as positive when at least two bands, including p24 and gp41 or gp 120/160, are present. The clinical importance of WBi patterns for HIV-1is unknown. Most studies on this issue have been done on blood donors positive on anti-HIV-1 enzyme immunoassay (EIA); 48-64% of donors repeatedly reactive for anti-HIV-1 by EIA have WBi patterns.12 The frequency of such patterns in low-risk populations is so high as to suggest that, as with EIA, most such reactions represent false-positive results. If WBi patterns were common among individuals nonreactive for anti-HIV-1 by EIA, it would favour the hypothesis that such WB patterns are related to nonspecificity of the assay, test artifact, or a common crossreacting antibody unrelated to HIV-1. We therefore measured the prevalence of WBi in randomly selected EIA-negative donors whose recipient sera were available from previous prospective studies on viral hepatitis. The availability of recipient sera allowed for an examination of HIV-1transmission by WBi donors. a test
METHODS
Study Design The donor-recipient population included in this study was selected from prospective studies of post-transfusion hepatitis conducted at the National Institutes of Health in 1984-87. 100 donors were initially tested for anti-HIV-1by EIA and WB. The donor population was chosen in such a way as to maximise the number of recipients followed up; two donors for each recipient were randomly selected from all the donors to each of 50 recipients. The 6-month post-transfusion samples of the recipients of blood from donors with a WBi result were then tested by EIA and WB. If these showed reactivity to EIA or WB, a pretransfusion sample and 1 and 3 month post-transfusion samples were then assayed. The entire group of 4 samples from each recipient was tested on two separate occasions by WB in the same laboratory, the second time under code. A control group of recipients who had received only WB negative blood was also investigated.
DONORS WITH AN
HIV antigen assay (Abbott Laboratories, North Chicago, Illinois, USA) and with an unlicensed recombinant immunoblot assay system (RIBA-HIV216, Chiron Corporation, Emeryville, California, USA) that contains four recombinant antigens (p24, p31, gp41, and gpl20) on an EIA strip. A recombinant EIA anti-HIV-2 test (Johnson & Johnson Bio. Center, La Jolla, California, USA) and an EIA anti-HTLV-I assay (Du Pont) were also used in a selected group of WBi samples.
Polymerase Chain Reaction DNA extracted from
serum
(0-4 ml) by treatment with sodium
dodecyl sulphate (SDS) and proteinase K was amplified by polymerase chain reaction (PCR). Total nucleic acid was resuspended in 100 ul of TE buffer (10 mmol/1 "tris" (pH 75), 1 mmol/1 edetic acid) and amplified in a thermal cycler (PerkinElmer Cetus, Emeryville, California, USA) by use of a commercially available reagent kit (Perkin-Elmer Cetus) according to manufacturer’s instructions. 20 ul of the DNA sample were amplified with primer pairs to the gag (SK 38/39) and env (SK 68/69) region of HIV-1 by co-amplification.4 Sequences of these primer pairs and respective probes have been previously published.5 Incubation conditions were denaturation at 94°C for 1 min, primer annealing at 55°C for 1 min, and primer extension at 65°C for 1 min 30 s for 35 cycles. Sensitivity and specificity for the assay have been previously described.6.7 In summary, in more than 500 serum or plasma samples, sensitivity of the PCR was 100% for anti-HIV-1 positive individuals with symptoms and 96% for
symptom-free individuals. RESULTS
20 of the 100 donors negative for anti-HIV by EIA showed a WBi pattern, whereas 80 were WB negative (table I). Most of the bands were reactive against the gag proteins, and p24 was the most frequently detected (70% of WBi donors), in 10 cases as a single band and in 4 cases in combination with other bands. The 20 units of WBi blood were transfused to 19 recipients. 7 (36%) of the 19 recipients had a WBi pattern in their 6 month post-transfusion sample (table II); this was similar to the finding of WBi in 8 (42%) of control recipients. Anti-gag antibodies were also predominant in the recipients, and there was no correlation between the WBi pattern of the donor and the WBi pattern of the corresponding recipient (table III). No recipient of WBi blood became WB positive by FDA criteria or developed anti-HIV by EIA. To assess reproducibility and consistency of WBi patterns, a pretransfusion sample and the 1,3, and 6 month TABLE II-ANTI-HIV STATUS OF RECIPIENTS OF WB
Serological Assays The anti-HIV-1 EIA
20
INDETERMINATE WB RESULT
INDETERMINATE AND WB NEGATIVE BLOOD
commercial test (Du Pont Co, Wilmington, Delaware, USA) and the WB test used was the only one licensed by the FDA (Biotech/Dupont HIV western blot kit). A WBi pattern was defined as the detection of any band or combination of bands that did not meet the FDA criteria for WB positivity (presence of antibodies against the env, gag, and pol products).3 Only WB bands with at least a reactivity score of 1 + were considered in this study. WBi samples were also tested with an was a
*1
recipient transfused with two units ofWB indeterminate blood.
1025 TABLE III-WB PATTERN OF THE 7 DONORS AND CORRESPONDING RECIPIENTS WITH WB INDETERMINATE RESULT 6 MONTHS POST-TRANSFUSION
post-transfusion samples of the 15 recipients who had a WBi result (7 recipients of WBi blood and 8 control recipients of WB negative blood) were simultaneously assayed by WB; in only 5 recipients was a consistent WBi pattern observed in the four serial samples. The remaining 10 recipients showed either different WBi patterns or alternately negative and indeterminate results in each of the serial samples. When WB analysis of the four serial samples from each of the 15 recipients was repeated under code, in only 21 of the 60 total samples (35%) was the initial WB pattern exactly reproducible in the repeat test. A p24 band was reproducible in only 25% of the samples in which it was detected at least once, whereas p55 was reproducible in 78%. None of the WBi samples from donors or recipients were reactive for HIV-1 antigen and only 1 of the WBi donors showed reactivity to p24 in the recombinant immunoblot assay; the rest of the WBi donors and recipients were negative in this assay. When a selected group of 5 donor-recipient pairs with the most consistent WBi patterns was tested for anti-HIV-2 and anti-HTLV-I by EIA, none was reactive for anti-HIV-2, but two samples from 1 recipient were repeatedly reactive for anti-HTLV-I. The WB results for HTLV-1 in these two samples were negative. PCR analysis of HIV-1DNA sequences by use of gag and env primers and of HTLV-1 DNA sequences by use of pol primers was negative in all donors and recipients. In two samples a very weak signal for gag products was initially obtained, but the result was not reproducible in subsequent experiments. DISCUSSION
Although all 100 donors selected were anti-HIV-1 EIA negative, 20% gave indeterminate WB patterns. This high prevalence of WBi results in an EIA-negative, low-risk donor population suggested that these reactions were non-specific. None of the WBi donors was HIV-1 antigen positive and all but
1
were
non-reactive to the recombinant
HIV-1 strips. Transmission data further supported the
non-specificity of a WBi result. None of the
19
recipients of
WBi blood became EIA positive over a 6-month interval and none became WB positive, tested positive for HIV-1 antigen, or reacted to the recombinant strips. However, a WBi pattern was detected in 36% of these recipients; this
probably represented the same form of false positivity rather than HIV-1 infection since 42% of controls, who received WB negative blood, were also WBi. Non-specificity was further emphasised by the finding that a sample, when retested, often showed a different indeterminate pattern or became WB negative, and that serial samples from the same individual had variable WB patterns
over
time without
progressing to a true-positive WB. In the 5 recipients with a consistent WBi, the pattern had already been established in the pretransfusion sample.
Absence of HIV-1 infection in these WBi donors and recipients was confirmed by negative reactions for HIV-1 related nucleic acid when samples were tested by PCR with probes for both gag and env regions. PCR was done on serum rather than cells because of non-availability of stored lymphocytes from the donor-recipient pairs; however, serum PCR had been shown to be quite sensitive.6,7 The reason for the high frequency of false-positive reactions is unclear but seems to be primarily an artifact of the assay. We did not find evidence of cross-reactivity with HIV-2 or HTLV-1, although such cross-reactivity has been reported.8 The existence of anti-HLA antibodies, particularly in the recipient group, might explain some WBi patterns,8 but is unlikely to account for many of these reactions. The lack of reactivity against recombinant HIV antigens suggests that the non-specificity is the product of a contaminant of the viral lysate used in the WB assay or some additional artifact of the procedure per se. The high frequency of WBi patterns in both EIA-positive and EIA-negative donors and the infectivity data presented here make it probable that most donors with WBi patterns do not harbour or transmit HIV-1. However, some studies have suggested that there is a small risk (2-5 %) of true HIV infection in EIA-positive WBi donors as indicated by the development of a positive WB within 2 or 3 months;1,S when these donors are interviewed, a history of high-risk behaviour is often obtained. It is clearly prudent to continue to exclude blood units which are EIA positive and WBi, and it is important to follow up the donors over time to ensure that they do not become fully positive WB or start to show other laboratory or clinical evidence of HIV-1 infection. However, the data in this study and others9 indicate that among WBi donors who have no history of exposure to HIV-1 and whose WBi pattern does not become WB positive in 6 months, the likelihood of HIV-1infection is remote; in such donors there is high probability that the WBi pattern represents a non-specific reaction that places neither the donor nor their contacts at risk of HIV-1 infection. Correspondence and reprint requests should be addressed to J. G., Department of Transfusion Medicine, Warren G. Magnuson Clinical Center, Bldg 10A, Rm IN309, National Institutes of Health, Bethesda, Maryland 20892, USA. REFERENCES 1. Kleinman S, Fitzpatrick L, Secord K, Wilke D. Follow-up testing and notification of anti-HIV western blot atypical (indeterminant) donors. Transfusion 1988; 28: 280-82. 2. Hosein B, Bianco C. Indeterminate western blot for HIV-1 does not suggest imminent seroconversion of blood donors. Proceedings of the IV International Conference on AIDS, Stockholm. 1988; book 2: 348. 3. Centers for Disease Control. Update: serologic testing for antibody to human immunodeficiency virus. MMWR 1988; 36: 833-45. 4. Hewlett IK, Ruta M, Cristiano K, Ann Hawthorne C, Epstein JS. Co-amplification of multiple regions of the HIV-1 genome of the polymerase chain reaction: potential use in multiple diagnosis. Oncogene (in press). 5. Ou C, Kwok S, Mitchell SW, et al. DNA amplification for direct detection of HIV-1 in DNA of peripheral blood mononuclear cells. Science 1988; 239: 295-97. 6. Hewlett IK, Ann Hawthorne C, Ruta M, et al. Sensitivity and specificity of HIV-1 gag and env gene detection in plasma and serum of HIV-infected individuals by the polymerase chain reaction. J Cell Biochem 1989; suppl 13E: 289. 7. Hewlett IK, Gregg RA, Ann Hawthorne C, et al. Detection of HIV-1-specific DNA and RNA in plasma by the polymerase chain reaction prior to seroconversion. In: Lerner RA, Ginsberg H, Chanock RM, Brown F, eds. Vaccines 89: modem approaches to new vaccines including prevention of AIDS. New York: Cold Spring Harbor Laboratory, 1989: 167-72 8. Dock NL, Lamberson HV, O’Bnen TA, Tribe DE, Alexander SS, Poiesz BJ. Evaluation of atypical human immunodeficiency virus immunoblot reactivity in blood donors. Transfusion 1988; 28: 412-18. 9. Van Der Poel CL, Lelie PN, Reesink HW, et al. Blood donors with indeterminate anti-p24 gag reactivity m HIV-1 Western blot: absence of infectivity to transfused patients and m virus culture. Vox Sang 1989; 56: 162-67.
prevalence periods.
of
urinary catheterisation
between the
two
Public Health DISCUSSION
We have shown that repeated prevalence surveys are feasible in a large general hospital, and that the results can be used to measure statistical trends in hospital infection. The British National Prevalence Survey protocol was simple to administer and use of a microcomputer for data correction and descriptive analysis allowed the production of summary statistics within a few weeks. We did not assess the reproducibility or accuracy of the method, but our results over 3 years with varying survey teams suggested that both were acceptable. This conclusion is supported by several fmdings: the risk-factors predisposing to HAI were typical 10,13 (including extremes of age, severity of underlying disease, acute or intensive medical service, long hospital stay, surgical operation, and urinary and intravenous catheterisation); the sites of HAI and the organisms involved have been reported elsewhere and were also typical; 10,21-23 and the prevalence rates of HAI and CAI were in the range expected for our patient population.6,10,23 CAI prevalence rates remained constant over the study period, but HAI rates fell after the introduction of infection control policies. We presumed that some of this reduction in HAI was due to infection control. Statistical analysis of the survey data showed that there was significant variation in risk-factors of HAI in different surveys. Analysis of the rates adjusted for risk-factors confirmed that the prevalence of HAI and HAUTI had fallen significantly after the introduction of appropriate infection control policies, independent of variations in risk. The 30% reduction in HAI following improvements in infection control in this prevalence study is similar to that found in the SENIC incidence study.6 Epidemiological criteria have been established to test a cause and effect relation.24 There should be a consistent relation between the cause and effect in different study populations; the size of the effect should be reasonably large; the strength of the association should be high, more exposure should lead to more effect; the effect should follow the intervention, and a reasonable model should exist to explain the effect.24 Freeman and McGovan4 comment that these criteria have rarely been applied to studies of infection control. The study reported here fulfils all these criteria except the first, but further work will show whether these results are repeatable in other populations. If they are, this method could be used to test the effectiveness of different infection control policies, perhaps with different hospitals or different wards acting as control sites. Repeated prevalence surveys offer a practical and sensitive way of measuring hospital infection and should be used more widely. They provide data for both infected and non-infected patients and allow adjustment of infection rates for risk. Prevalence surveys can be used to assess the impact of infection control programmes on hospital acquired infection. We believe this technique is appropriate for the analysis of hospital infection, and is especially useful in hospitals with limited resources. We
are
grateful
to
the medical and
nursing staff of the
Prince of Wales
Hospital who assisted in the surveys, and especially to our infection control Alice Chow, Christiana Chan, Deborah Ho, and Rosita Kwok. We thank Corrie Ng and Amy Cheung for computer data entry, and Linda Ip for nurses
secretarial assistance.
WHAT DO WESTERN BLOT INDETERMINATE PATTERNS FOR HUMAN IMMUNODEFICIENCY VIRUS MEAN IN EIA-NEGATIVE BLOOD DONORS?
JOAN GENESCA1 BETSY W. JETT1 JAY S. EPSTEIN2
J. WAI-KUO SHIH1 INDIRA K. HEWLETT2 HARVEY J. ALTER1
Department of Transfusion Medicine, Warren G. Magnuson Clinical Center, National Institutes of Health,1 and Division of Blood and Blood Products, Food and Drug Administration,2 Bethesda, Maryland, USA To investigate the specificity of western blot indeterminate (WBi) patterns for antibodies to the human immunodeficiency virus type 1 (HIV-1), 100 enzyme immunoassay (EIA) negative donors from whom prospectively obtained recipient sera were available were tested by WB. 20 were WBi, with p24 being the predominant (70%) and generally the only band. Among recipients of WBi blood, 36% were WBi in their 6 month post-transfusion sample, but so were 42% of a control population that had received only WB-negative
Summary
G. L. FRENCH AND OTHERS: REFERENCES 1.
2.
Haley RW, Schaberg DR, Crossley KB, von Allmen SD, McGowan JE. Extra charges and prolongation of stay attributable to nosocomial infections: a prospective interhospital comparison. Am J Med 1981; 70: 51-57. Brachmann PS. Nosocomial infection control: an overview. Rev Infect Dis 1981; 3: 640-48.
3. Daschner F. Economic aspects of hospital infections. J Hosp Infect 1982; 3: 1-4. 4. Freeman J, McGowan JE. Methodologic issues in hospital epidemiology. I. Rates, case-finding and interpretation. Rev Infect Dis 1981; 3: 658-67. 5. Tager IB, Ginsberg MB, Simchen E, Miao L, Holbrook K, Faich GA. Rationale and methods for a statewide, prospective surveillance system for the identification and prevention of nosocomical infections. Rev Infect Dis 1981; 3: 683-93. 6. Haley RW, Culber DH, White JW, et al. The efficacy of infection surveillance and control programs in preventing nosocomial infections in US hospitals. Amer J
Epidemiol 1985, 121: 182-205. Infect 1980; 1: 293-97. hospitals. Hosp J J, McGowan JE. Methodologic issues in hospital epidemiology. II. Time and accuracy in estimation. Rev Infect Dis 1981; 3: 668-77. 9. Freeman J, Hutchinson GB Prevalence, incidence and duration. Amer J Epidemiol 1980; 112: 707-23. 10. Meers PD, Ayliffe GAJ, Emmerson AM, et al. Report on the national survey of infection in hospitals, 1980. J Hosp Infect 1981; 2: (supplement). 11. Rhame FS, Sudderth WD. Incidence and prevalence as used in the analysis of the occurrence of nosocomial infections. Amer Epidemiol 1981, 113: 1-11. J 12. Aber RC, Bennett JV. Surveillance of nosocomial infection In: Bennett JV, Brachman PS, eds. Hospital infections. Boston: Little, Brown, 1979: 53-61. 13. Freeman J, McGowan JE. Risk factors of hospital infection. J Infect Dis 1973; 138: 7. Casewell MW Surveillance of infection in
8. Freeman
811-19. 14. Storch GA, Rajagopolan L. Methicillin-resistant
Staphylococcus aureus bacteraemia in children. Pediatr Infect Dis 1986, 5: 59-67 15. French GL, Cheng A, Farrington M. Prevalence survey of infection in a Hong Kong hospital using a standard protocol and microcomputer data analysis. J Hosp Infect 1987; 9: 132-42. Lowbury EJL, Ayliffe CAJ, Geddes AM, Williams JD Control of Hospital Infection London: Chapman & Hall; 1975 17. Wong ES. Guideline for prevention of catheter-associated urinary tract infections. Amer J Infect Control 1983; 11: 28-33. 18 Kahn HA. An introduction to epidemiological methods New York Oxford University Press, 1983 19. Dixon WJ, ed BMDP statistical software. Berkley: University of California Press. 16.
1985.
Armitage P. Statistical methods in medical research. Oxford: Blackwell, 1971. 21. French GL, Cheng AFB Prevalence of hospital infection at the Prince of Wales Hospital. J Hong Kong Med Assoc 1989; in press. 22 Marples RR, Mackintosh CA, Meers PD. Microbiological aspects of the 1980 national prevalence survey of infections in hospitals. J Hosp Infect 1984; 5: 172-80. 23. Haley RW, Culber DH, White JW, Morgan WM, Emori TG. The nationwide nosocomial infection rate. a new need for vital statistics. Amer J Epidemiol 1985; 20.
121: 159-67. 24. Hill AB. Principles of medical statistics. 9th ed. New York Oxford 1971, p 390
University Press:
1024 TABLE I-WB PATTERNS OF THE
blood. When serial samples from recipients with a WBi pattern were tested on two occasions, only 35% of results were reproducible. No recipient of WBi blood became EIA positive, true positive for WB, positive for HIV-1 antigen, or positive for EIA reactivity against recombinant p24 or gp41. The polymerase chain reaction was negative for gag and env HIV-1 sequences in all donors and recipients. Thus WBi patterns are exceedingly common in randomly selected donors and recipients and such patterns do not correlate with the presence of HIV-1 or the transmission of HIV-1 from donor to recipient. INTRODUCTION
WHEN the western blot (WB) assay was licensed for use as for human immunodeficiency virus type-1 (HIV-1), the criteria laid down for a positive result in the absence of risk factors or clinical suspicion were the presence of protein bands at the gag, env, and pol positions. Any pattern of bands which lacks this full array is considered WB indeterminate (WBi), although some experts regard WB as positive when at least two bands, including p24 and gp41 or gp 120/160, are present. The clinical importance of WBi patterns for HIV-1is unknown. Most studies on this issue have been done on blood donors positive on anti-HIV-1 enzyme immunoassay (EIA); 48-64% of donors repeatedly reactive for anti-HIV-1 by EIA have WBi patterns.12 The frequency of such patterns in low-risk populations is so high as to suggest that, as with EIA, most such reactions represent false-positive results. If WBi patterns were common among individuals nonreactive for anti-HIV-1 by EIA, it would favour the hypothesis that such WB patterns are related to nonspecificity of the assay, test artifact, or a common crossreacting antibody unrelated to HIV-1. We therefore measured the prevalence of WBi in randomly selected EIA-negative donors whose recipient sera were available from previous prospective studies on viral hepatitis. The availability of recipient sera allowed for an examination of HIV-1transmission by WBi donors. a test
METHODS
Study Design The donor-recipient population included in this study was selected from prospective studies of post-transfusion hepatitis conducted at the National Institutes of Health in 1984-87. 100 donors were initially tested for anti-HIV-1by EIA and WB. The donor population was chosen in such a way as to maximise the number of recipients followed up; two donors for each recipient were randomly selected from all the donors to each of 50 recipients. The 6-month post-transfusion samples of the recipients of blood from donors with a WBi result were then tested by EIA and WB. If these showed reactivity to EIA or WB, a pretransfusion sample and 1 and 3 month post-transfusion samples were then assayed. The entire group of 4 samples from each recipient was tested on two separate occasions by WB in the same laboratory, the second time under code. A control group of recipients who had received only WB negative blood was also investigated.
DONORS WITH AN
HIV antigen assay (Abbott Laboratories, North Chicago, Illinois, USA) and with an unlicensed recombinant immunoblot assay system (RIBA-HIV216, Chiron Corporation, Emeryville, California, USA) that contains four recombinant antigens (p24, p31, gp41, and gpl20) on an EIA strip. A recombinant EIA anti-HIV-2 test (Johnson & Johnson Bio. Center, La Jolla, California, USA) and an EIA anti-HTLV-I assay (Du Pont) were also used in a selected group of WBi samples.
Polymerase Chain Reaction DNA extracted from
serum
(0-4 ml) by treatment with sodium
dodecyl sulphate (SDS) and proteinase K was amplified by polymerase chain reaction (PCR). Total nucleic acid was resuspended in 100 ul of TE buffer (10 mmol/1 "tris" (pH 75), 1 mmol/1 edetic acid) and amplified in a thermal cycler (PerkinElmer Cetus, Emeryville, California, USA) by use of a commercially available reagent kit (Perkin-Elmer Cetus) according to manufacturer’s instructions. 20 ul of the DNA sample were amplified with primer pairs to the gag (SK 38/39) and env (SK 68/69) region of HIV-1 by co-amplification.4 Sequences of these primer pairs and respective probes have been previously published.5 Incubation conditions were denaturation at 94°C for 1 min, primer annealing at 55°C for 1 min, and primer extension at 65°C for 1 min 30 s for 35 cycles. Sensitivity and specificity for the assay have been previously described.6.7 In summary, in more than 500 serum or plasma samples, sensitivity of the PCR was 100% for anti-HIV-1 positive individuals with symptoms and 96% for
symptom-free individuals. RESULTS
20 of the 100 donors negative for anti-HIV by EIA showed a WBi pattern, whereas 80 were WB negative (table I). Most of the bands were reactive against the gag proteins, and p24 was the most frequently detected (70% of WBi donors), in 10 cases as a single band and in 4 cases in combination with other bands. The 20 units of WBi blood were transfused to 19 recipients. 7 (36%) of the 19 recipients had a WBi pattern in their 6 month post-transfusion sample (table II); this was similar to the finding of WBi in 8 (42%) of control recipients. Anti-gag antibodies were also predominant in the recipients, and there was no correlation between the WBi pattern of the donor and the WBi pattern of the corresponding recipient (table III). No recipient of WBi blood became WB positive by FDA criteria or developed anti-HIV by EIA. To assess reproducibility and consistency of WBi patterns, a pretransfusion sample and the 1,3, and 6 month TABLE II-ANTI-HIV STATUS OF RECIPIENTS OF WB
Serological Assays The anti-HIV-1 EIA
20
INDETERMINATE WB RESULT
INDETERMINATE AND WB NEGATIVE BLOOD
commercial test (Du Pont Co, Wilmington, Delaware, USA) and the WB test used was the only one licensed by the FDA (Biotech/Dupont HIV western blot kit). A WBi pattern was defined as the detection of any band or combination of bands that did not meet the FDA criteria for WB positivity (presence of antibodies against the env, gag, and pol products).3 Only WB bands with at least a reactivity score of 1 + were considered in this study. WBi samples were also tested with an was a
*1
recipient transfused with two units ofWB indeterminate blood.
1025 TABLE III-WB PATTERN OF THE 7 DONORS AND CORRESPONDING RECIPIENTS WITH WB INDETERMINATE RESULT 6 MONTHS POST-TRANSFUSION
post-transfusion samples of the 15 recipients who had a WBi result (7 recipients of WBi blood and 8 control recipients of WB negative blood) were simultaneously assayed by WB; in only 5 recipients was a consistent WBi pattern observed in the four serial samples. The remaining 10 recipients showed either different WBi patterns or alternately negative and indeterminate results in each of the serial samples. When WB analysis of the four serial samples from each of the 15 recipients was repeated under code, in only 21 of the 60 total samples (35%) was the initial WB pattern exactly reproducible in the repeat test. A p24 band was reproducible in only 25% of the samples in which it was detected at least once, whereas p55 was reproducible in 78%. None of the WBi samples from donors or recipients were reactive for HIV-1 antigen and only 1 of the WBi donors showed reactivity to p24 in the recombinant immunoblot assay; the rest of the WBi donors and recipients were negative in this assay. When a selected group of 5 donor-recipient pairs with the most consistent WBi patterns was tested for anti-HIV-2 and anti-HTLV-I by EIA, none was reactive for anti-HIV-2, but two samples from 1 recipient were repeatedly reactive for anti-HTLV-I. The WB results for HTLV-1 in these two samples were negative. PCR analysis of HIV-1DNA sequences by use of gag and env primers and of HTLV-1 DNA sequences by use of pol primers was negative in all donors and recipients. In two samples a very weak signal for gag products was initially obtained, but the result was not reproducible in subsequent experiments. DISCUSSION
Although all 100 donors selected were anti-HIV-1 EIA negative, 20% gave indeterminate WB patterns. This high prevalence of WBi results in an EIA-negative, low-risk donor population suggested that these reactions were non-specific. None of the WBi donors was HIV-1 antigen positive and all but
1
were
non-reactive to the recombinant
HIV-1 strips. Transmission data further supported the
non-specificity of a WBi result. None of the
19
recipients of
WBi blood became EIA positive over a 6-month interval and none became WB positive, tested positive for HIV-1 antigen, or reacted to the recombinant strips. However, a WBi pattern was detected in 36% of these recipients; this
probably represented the same form of false positivity rather than HIV-1 infection since 42% of controls, who received WB negative blood, were also WBi. Non-specificity was further emphasised by the finding that a sample, when retested, often showed a different indeterminate pattern or became WB negative, and that serial samples from the same individual had variable WB patterns
over
time without
progressing to a true-positive WB. In the 5 recipients with a consistent WBi, the pattern had already been established in the pretransfusion sample.
Absence of HIV-1 infection in these WBi donors and recipients was confirmed by negative reactions for HIV-1 related nucleic acid when samples were tested by PCR with probes for both gag and env regions. PCR was done on serum rather than cells because of non-availability of stored lymphocytes from the donor-recipient pairs; however, serum PCR had been shown to be quite sensitive.6,7 The reason for the high frequency of false-positive reactions is unclear but seems to be primarily an artifact of the assay. We did not find evidence of cross-reactivity with HIV-2 or HTLV-1, although such cross-reactivity has been reported.8 The existence of anti-HLA antibodies, particularly in the recipient group, might explain some WBi patterns,8 but is unlikely to account for many of these reactions. The lack of reactivity against recombinant HIV antigens suggests that the non-specificity is the product of a contaminant of the viral lysate used in the WB assay or some additional artifact of the procedure per se. The high frequency of WBi patterns in both EIA-positive and EIA-negative donors and the infectivity data presented here make it probable that most donors with WBi patterns do not harbour or transmit HIV-1. However, some studies have suggested that there is a small risk (2-5 %) of true HIV infection in EIA-positive WBi donors as indicated by the development of a positive WB within 2 or 3 months;1,S when these donors are interviewed, a history of high-risk behaviour is often obtained. It is clearly prudent to continue to exclude blood units which are EIA positive and WBi, and it is important to follow up the donors over time to ensure that they do not become fully positive WB or start to show other laboratory or clinical evidence of HIV-1 infection. However, the data in this study and others9 indicate that among WBi donors who have no history of exposure to HIV-1 and whose WBi pattern does not become WB positive in 6 months, the likelihood of HIV-1infection is remote; in such donors there is high probability that the WBi pattern represents a non-specific reaction that places neither the donor nor their contacts at risk of HIV-1 infection. Correspondence and reprint requests should be addressed to J. G., Department of Transfusion Medicine, Warren G. Magnuson Clinical Center, Bldg 10A, Rm IN309, National Institutes of Health, Bethesda, Maryland 20892, USA. REFERENCES 1. Kleinman S, Fitzpatrick L, Secord K, Wilke D. Follow-up testing and notification of anti-HIV western blot atypical (indeterminant) donors. Transfusion 1988; 28: 280-82. 2. Hosein B, Bianco C. Indeterminate western blot for HIV-1 does not suggest imminent seroconversion of blood donors. Proceedings of the IV International Conference on AIDS, Stockholm. 1988; book 2: 348. 3. Centers for Disease Control. Update: serologic testing for antibody to human immunodeficiency virus. MMWR 1988; 36: 833-45. 4. Hewlett IK, Ruta M, Cristiano K, Ann Hawthorne C, Epstein JS. Co-amplification of multiple regions of the HIV-1 genome of the polymerase chain reaction: potential use in multiple diagnosis. Oncogene (in press). 5. Ou C, Kwok S, Mitchell SW, et al. DNA amplification for direct detection of HIV-1 in DNA of peripheral blood mononuclear cells. Science 1988; 239: 295-97. 6. Hewlett IK, Ann Hawthorne C, Ruta M, et al. Sensitivity and specificity of HIV-1 gag and env gene detection in plasma and serum of HIV-infected individuals by the polymerase chain reaction. J Cell Biochem 1989; suppl 13E: 289. 7. Hewlett IK, Gregg RA, Ann Hawthorne C, et al. Detection of HIV-1-specific DNA and RNA in plasma by the polymerase chain reaction prior to seroconversion. In: Lerner RA, Ginsberg H, Chanock RM, Brown F, eds. Vaccines 89: modem approaches to new vaccines including prevention of AIDS. New York: Cold Spring Harbor Laboratory, 1989: 167-72 8. Dock NL, Lamberson HV, O’Bnen TA, Tribe DE, Alexander SS, Poiesz BJ. Evaluation of atypical human immunodeficiency virus immunoblot reactivity in blood donors. Transfusion 1988; 28: 412-18. 9. Van Der Poel CL, Lelie PN, Reesink HW, et al. Blood donors with indeterminate anti-p24 gag reactivity m HIV-1 Western blot: absence of infectivity to transfused patients and m virus culture. Vox Sang 1989; 56: 162-67.