Hepatitis C Virus Prevalence and Clearance Among U.S. Blood Donors, 2006-2007: Associations with Birth Cohort, Multiple Pregnancies and Body Mass Index

Abstract

Background

In 1992-1993, the prevalence of anti-HCV among U.S. blood donors was 0.36 percent, but contemporary data on antibody and RNA prevalence are lacking.

Methods

We performed a large, cross-sectional study of blood donors at six U.S. blood centers in 2006-2007. Anti-HCV was measured by EIA followed by immunoblot, and HCV RNA by nucleic acid testing. Adjusted odds ratios (aORs) were derived using multivariable logistic regression.

Results

Among 959,281 donors, 695 had anti-HCV (prevalence 0.072%), of whom 516 (74%) were RNA+ and 179 (26%) RNA-. Compared to 1992-1993, prevalence was lower and peaked in older age groups. Anti-HCV was associated with body mass index (BMI) >30 kg/m² (aOR=0.6, 95% CI 0.5-0.8) and among women, higher gravidity (aOR for >= 5 vs. 0 pregnancies = 3.2, 95% CI 1.9-5.4). HCV RNA negative status was associated with Black race (aOR=0.4, 95% CI 0.2-0.7), more than high-school education (aOR=1.6, 95% CI 1.1-2.4) and BMI >30 (aOR=2.4, 95% CI 1.4-3.9).

Conclusions

Declining HCV prevalence is most likely due to culling of seropositive donors and a birth cohort effect. We found new associations between Anti-HCV prevalence and gravidity and obesity. Recently discovered genetic factors may underlie differences in HCV RNA clearance among Black donors.

Background

Hepatitis C virus (HCV) is a blood-borne virus that causes chronic hepatitis, cirrhosis and hepatocellular carcinoma and has recently been implicated in the pathogenesis of non-Hodgkin's lymphoma[1, 2]. More than 170 million people worldwide are anti-HCV+, with prevalences levels generally of 1% to 2% in most populations, but occasionally at much higher levels due to instances of iatrogenic infection. In the United States prevalence is 1.6% in the general population with highest prevalence in males and those aged 40 to 49 years[3]. In United States blood donors, prevalence was estimated at 0.36% in the early 1990s, again with highest prevalence in males and middle-aged donors[4]. One study showed a significant decline in HCV prevalence among first-time U.S. blood donors from 0.63% in 1991 to 0.40% in1996 whereas another showed overall decline in HCV prevalence but an increase among 50 to 59 year old men[5, 6]. The incidence among repeat blood donors is low, and since the institution of nucleic acid testing (NAT) for HCV RNA in 1999 allowing the discovery and quarantine of early antibody-negative “window period” blood units, the residual risk of HCV infection has been estimated to be one in 1.8 million blood units transfused[7].

However, recent data on HCV prevalence and risk factors among a large sample of U.S. blood donors are lacking. Such data are important in order to understand the burden of infection in different donor subgroups and to predict future prevalence for public health purposes. From an epidemiologic standpoint, it has been postulated that HCV infection in the United States demonstrates a birth cohort effect due to an epidemic of transmission by injection drug use in the 1960s and 1970s[4]. If such an effect could be proven by showing reduced prevalence in more contemporary data, it would have important implications for modeling the future burden of HCV infection in the United States and the health care impact of HCV related disease outcomes. Current parallel testing of blood donors for anti-HCV and RNA also allows the investigation of the determinants of presumptively resolved HCV infection as manifested by antibody positive/RNA negative status.

We therefore performed a large, cross-sectional prevalence study of HCV prevalence among blood donors at a research network of six blood centers throughout the United States. Data were compared with those from an earlier published study from a similar U.S. blood center network[4], and a separate analysis of the determinants of RNA+/RNA- status was undertaken.

Methods

Study population and design

This was a cross-sectional seroprevalence study among blood donors at six U.S. blood centers from Jan 2006 thru Sept 2007. The blood centers participated in the Retrovirus Epidemiology Donor Study-II (REDS-II), and included American Red Cross Blood Services, New England (Vermont, New Hampshire, Maine and Massachusetts) and Southern (Georgia) regions, the Institute for Transfusion Medicine (Pennsylvania), the Hoxworth Blood Center (Ohio), the Blood Center of Wisconsin (Wisconsin) and Blood Centers of the Pacific (California). Computerized data on all blood donations during the study period was forwarded to the coordinating center and merged into a central research database. Each blood donor was represented only once in the data presented here; the donor was considered positive if any donation during the study period tested positive. First-time and repeat donors had or did not have, respectively, record of a previous donation at that center. Body mass index (BMI) was categorized as normal (under 25 kg/m²), overweight (25 to 29.99 kg/m²) or obese (30 or more kg/m²). All anti-HCV positive donors were notified and excluded from future donation according to blood center operational protocols. Data collection was carried out under protocols approved by institutional review boards at the six participating blood centers.

HCV testing

For the prevalence analysis, HCV infection was defined as the presence of anti-HCV (positive enzyme immunoassay followed by positive recombinant immunoblot). For the analysis of chronic versus resolved HCV infection, the results from parallel NAT was used to sort anti-HCV positives into persistently viremic (HCVRNA+) and presumptively resolved or resolving (HCV RNA-) HCV infections. All laboratory testing was done using Food and Drug Administration-licensed assays and under strict standard operating procedures at the blood center laboratories. Anti-HCV was measured with third generation enzyme immunoassays (EIA); recombinant immunoblot (RIBA) confirmation was done either on all EIA positive samples (5 centers) or on EIA positive samples that were NAT negative (1 center) and this had no effect on proportions of confirmed anti-HCV between centers. NAT was done with either transcription mediated amplification (Genprobe/Novartis; pools of 16) or polymerase chain reaction (Roche Laboratories; pools of 24) techniques.

Statistical analysis

We calculated overall and subgroup-specific anti-HCV and HCV RNA prevalences with 95 percent confidence intervals (95% CI) using the Clopper-Pearson method [8]. Distributions of BMI, defined as weight in kilograms divided by height in meters squared, were compared between HCV status groups using ANOVA. Three separate multivariable logistic regression models were constructed and used to calculate adjusted odds ratios (aOR's) and 95% CI's: i) the prevalence analysis included all HCV-screened donors and tested associations between anti-HCV positivity and donor characteristics, irrespective of RNA status; ii) a second prevalence model examining gravidity was similar to the former one but restricted to females only; and iii) the analysis of resolved versus chronic HCV infection included anti-HCV positives only and tested associations of HCV RNA negative status with various donor characteristics. We used a backward elimination strategy in constructing each multivariable model. All variables were included in the initial model; those without associations (p>0.10) were then excluded from the final model unless forced in because of biological plausibility or previously published findings.

Results

We studied 959,281 donors of whom 695 (72 per 10⁵) had evidence of HCV infection, including 516 (74%) with both anti-HCV and RNA and 179 (26%) with anti-HCV only. Subgroup-specific prevalence and aOR's are presented in Table 1. Anti-HCV prevalence was much higher in the first-time than repeat donors, and was higher in male than female donors. Age-specific prevalence increased to a maximum in the 40-49 and 50-59 year-old age groups, and declined thereafter (Figure 1B). Prevalence was lower in 2006-07 than in previously published 1992-1993 data in all age and sex groups (Figure 1A)[4], and peak prevalence shifted to older age groups. Prevalence was highest among those of African American race/ethnicity, similar in non-Hispanic Whites and Hispanics, and lower among Asians.

TABLE 1.

Prevalence of anti-HCV in various donor groups and adjusted odds ratio for the association with each variable. Reference categories are indicated by adjusted odds ratio (aOR) = 1.0; significant associations are shown in bold font.

Donor Group	N Donors	n anti-HCV Positive	Prevalence per 105	aOR (95% CI)
Donor Status
First time donor	280,612	631	225	38.3 (29.1-50.4)
Repeat donor	675,294	57	8	1.0 -----
Age
<20	179,427	45	25	0.1 (0.1-0.2)
20-29	158,617	82	52	1.0 ---
30-39	141,220	102	72	2.0 (1.5-2.7)
40-49	199,592	246	123	4.3 (3.3-5.5)
50-59	178,893	191	107	4.7 (3.6-6.2)
60+	101,526	29	29	1.2 (0.8-1.9)
Sex
Male	467,996	415	89	1.8 (1.6-2.1)
Female	491,255	279	57	1.0 -----
Race/Ethnicity
White	810,716	531	65	1.0 -----
Black	56,467	78	138	1.4 (1.1-1.8)
Hispanic	34,722	24	69	0.7 (0.5-1.1)
Asian	22,351	8	36	0.5 (0.2-0.9)
Education
High School or less	265,874	312	117	8.1 (6.2-10.5)
Some College	281,132	219	78	3.8 (2.9-4.9)
College graduate	327,961	74	23	1.0 -----
Ever Transfused
No	815,780	503	62	1.0 -----
Yes	31,252	54	173	2.1 (1.6-2.8)
Body Mass Index
Underweight/Normal <25	342,961	213	62	1.0 -----
Overweight <=30	314,378	240	76	0.9 (0.7-1.1)
Obese >30	218,965	156	71	0.6 (0.5-0.8)
Blood Center
Wisconsin	122,071	48	39	1.0 -----
California	90,462	78	86	3.6 (2.3-4.9)
Georgia	250,759	234	93	2.8 (2.0-3.9)
Ohio	70,521	34	48	2.2 (1.4-3.3)
MA/VT/NH/ME	303,740	202	67	2.5 (1.8-3.4)
Pennsylvania	121,728	99	81	2.8 (2.0-4.0)

Figure 1.

Prevalence of HCV antibody among U.S. blood donors, REDS 1992-93 (Panel A, data from Murphy et al. JAMA 1996) and current data from 2006-07 (Panel B).

In the multivariable logistic regression model for anti-HCV prevalence (Table 1), we found significant, independent associations between anti-HCV positivity and middle age, male sex and African-American vs. non-Hispanic White race/ethnicity. Donors with Asian race/ethnicity were significantly less likely to be anti-HCV positive. Significant associations were also seen with high school or lower education, previous blood transfusion, blood center and, inversely, BMI.

Among female donors only, anti-HCV prevalence increased with the number of pregnancies: 30 per 100,000 (57/192,518) in nulligravida women; 57 per 100,000 (27/47,523) in women with one pregnancy; 71 per 100,000 (122/171,795 in women with two to four pregnancies; and 144 per 100,000 (29/20,072) in women with five or more pregnancies. In a separate multivariable model with female donors only and including previous transfusion and race/ethnicity along with age, body mass index, educational attainment, blood center and first-time versus repeat donor status, women with multiple pregnancies were significantly more likely to test anti-HCV positive (aOR for one pregnancy = 1.3, 95% CI 0.8-2.1; two to four pregnancies = 1.6, 95% CI 1.1-2.4; and five or more pregnancies = 3.2, 95% CI 1.9-5.5; all versus nulligravida women). Country of birth and blood donation procedure were eliminated from the final model due to lack of significance. Further analyses excluding repeat donors did not substantially change the results reported above.

We next evaluated associations with HCV RNA status among anti-HCV positives (Table 2). A total of 516 anti-HCV positives had HCV RNA and 179 did not; an additional 24 donors with incident HCV infection (HCV RNA positive but negative anti-HCV) were excluded from further analysis. HCV RNA negative status was less likely to be detected among donors of African-American race/ethnicity. However HCV RNA negative status was positively associated with greater than high school education, obesity defined as a BMI of greater than 30 and donation at the Georgia blood center. Female sex was not significantly associated with HCV RNA status (aOR=1.09, 95% CI 0.74-1.60) but it and other covariates were retained in the model because of prior reports in the literature and/or concern regarding potential confounding factors. Further analyses excluding repeat donors did not substantially change the results reported above.

Table 2.

Association of resolved HCV infection (anti-HCV without HCV RNA) with various donor characteristics, among anti-HCV positives only (N=695). Numbers may not add to totals due to missing values for some variables. Reference categories are indicated by adjusted odds ratio (aOR) = 1.0; significant associations are shown in bold font.

Variable	Antibody Positive (N=695)	Number (%) RNA- (N=179)	aOR	95% CI
Age < 20	45	6 (13)	1.00	---
20-29	82	20 (24)	1.44	0.5-4.1
30-39	102	29 (28)	1.58	0.6-4.4
40-49	246	60 (24)	1.49	0.6-3.9
50+	220	64 (29)	2.03	0.8-5.4
Male	415	102 (25)	1.00	---
Female	279	76 (27)	1.09	0.7-1.6
White race/ethnicity	531	146 (27)	1.00	---
African-American	78	13 (17)	0.37	0.2-0.7
Other race/ethnicity	131	29 (22)	1.39	0.7-2.8
Country Birth USA	591	156 (26)	1.00	---
Non-USA	26	6 (23)	0.89	0.3-2.5
High school or lower education	312	68 (22)	1.00	---
Some college or higher	293	89 (30)	1.61	1.1-2.4
Center: Wisconsin	48	9 (19)	1.00	---
California	78	17 (22)	0.96	0.4-2.5
Georgia	234	76 (32)	2.71	1.2-6.1
Ohio	34	7 (21)	1.07	0.3-3.3
MA / VT / NH / ME	202	43 (21)	1.07	0.5-2.4
Pennsylvania	99	27 (27)	1.69	0.7-4.1
First time donor	631	158 (25)	1.00	---
Repeat donor	57	19 (33)	1.28	0.7-2.4
Whole blood donation	660	168 (25)	1.00	---
Other procedure	35	11 (31)	1.24	0.6-2.8
Never transfused	503	132 (26)	1.00	---
Ever transfused	54	15 (28)	0.95	0.5-1.9
BMI Underweight/Normal (<25 kg/m2)	213	42 (20)	1.00	--
Overweight (25 to <30 kg/m2)	240	61 (25)	1.36	0.8-2.2
Obese (>=30 kg/m2)	156	56 (36)	2.36	1.4-3.9

Because of the above-noted inverse associations of BMI with anti-HCV and RNA, a graph of the frequency distribution of BMI according to anti-HCV and RNA status is shown in Figure 2. It shows that the entire distribution of BMI is shifted significantly to the right in antibody+/RNA- donors compared to both antibody+/RNA+ and double-negative donors (ANOVA p =0.0084). Antibody+/RNA- donors are both less likely to be thin and more likely to be overweight or obese than the other two groups.

Figure 2.

Distribution of body mass index according to HCV infection status. Subjects with presumptive resolved HCV (antibody+/NAT-; blue line) are compared to chronically infected (antibody+/NAT+; red line) and never infected (antibody-/NAT-; green line) subjects

Discussion

Compared to results from a decade and a half earlier, this study provides additional evidence that anti-HCV prevalence is declining among U.S. blood donors, most likely due to culling of seropositives as well as a birth cohort effect. In addition, we describe new associations of anti-HCV prevalence with increasing number of pregnancies among women, and with lower BMI among all donors. Finally, HCV RNA negative status was less likely to be observed among African-American and more likely to be observed among better educated and obese donors.

The decline in HCV prevalence in the current analysis compared to a similar study in 1992-1993[4] is consistent with that seen in other studies of blood donors in the U.S. and Canada[5, 6, 9]. Over this time, there has been little change in donor eligibility criteria, if any. We and others have previously proposed a birth cohort effect for HCV infection among North American blood donors as explaining declining prevalence, namely donors in the birth cohort at highest risk for HCV infection (those born from the late 1940's through the early 1960's) with the highest lifetime prevalence of injection drug use[10] are progressively less represented among active blood donors[4, 6, 9]. The current data showing lower age-specific prevalence in all age groups together with a shift in the maximum prevalence to a decade older donors compared with the 1992-1993 data supports the birth cohort hypothesis, as well as culling of seropositive donors since the introduction of HCV testing in the early 1990's. Similar indications of a birth cohort effect may be seen in U.S. and Japanese general population data[3, 11, 12]. Confirmation of a birth cohort effect is important because it implies that HCV prevalence in the U.S. will likely continue to decline over upcoming decades. The generation born in the late 1940's through early 1960's will likely manifest the peak incidence of HCV disease outcomes of cirrhosis and hepatocellular carcinoma, with declining rates in subsequent generations.

Our finding of an association of HCV prevalence with increasing gravidity is novel and of potential public health concern if substantiated by other studies. In favor of the validity of this association is the apparent dose-effect with increasing number of pregnancies, and its persistence despite adjustment for potential confounding by age, previous transfusion history, race/ethnicity and education. In light of the weight of evidence against sexual intercourse as a major transmission route of HCV, we do not believe the pregnancy association is a surrogate for sexual transmission[13]. Antenatal care and childbirth are accompanied by medical injections, instrumentation and a high proportion of caesarian sections that could be opportunities for iatrogenic transmission of HCV. Women in our study could also have been transfused during operations or anesthesia without remembering the event. The hypothesis of iatrogenic exposure is supported by the findings of a case control study of HCV infection among pregnant women in Karachi Pakistan, which found adjusted odds ratios of 1.25 for 3-4 pregnancies and 1.99 for 5 or more pregnancies. Those authors concluded that iatrogenic exposures during delivery could be the source of infection[14].

We could find no published U.S. studies to suggest iatrogenic transmission in the obstetric setting, although a large outbreak of HCV infection has recently been attributed to unsafe injection practices at a Las Vegas endoscopy clinic[15]. A review article counted 33 outbreaks in nonhospital health care settings in the past decade: 12 in outpatient clinics, 6 in hemodialysis centers, and 15 in long-term care facilities, resulting in 448 persons acquiring HBV or HCV infection[16]. Thus the hypothesis of a low-level iatrogenic risk may be plausible, and ought to be examined by other studies. Finally, in this cross-sectional study, we cannot rule out residual confounding, perhaps by age or socioeconomic status, as contributing to the observed association.

Our analysis of HCV nucleic acid status found that presumptive clearance of HCV RNA was less likely in African-American subjects and more likely in those with higher educational status and obesity; no significant association was found between the sexes in the multivariable analysis. The association with African-American race/ethnicity is consistent with a number of reports in the literature, and may be due to genetic differences in HLA type or interleukin 18 promoter polymorphism[1, 17, 18]. Polymorphism in the IL28b gene has recently been proposed as a major determinant of racial differences in both treatment response and spontaneous clearance of HCV viremia[19-22]. The association of HCV clearance with higher educational status was not found in the previous cross-sectional analysis among blood donors[23]. Better access to healthcare and antiviral treatment among higher socioeconomic status donors seems unlikely to explain this effect, since most donors were presumably unaware of their HCV status at the time of donation. Our lack of an association of female sex with HCV clearance adds to a divided literature on this topic. Several studies with and without adjustment for potential confounding factors have found that anti-HCV positive women are more likely to be RNA negative than their male counterparts[24-27], and a prospective study of young injection drug users observed a higher incidence of clearance among females[28]. However three other studies including subjects who were U.S. injection drug users, U.S. blood donors and Italian general population found no significant association of HCV clearance with sex[1, 5, 29, 30]. Because the latter studies generally showed a reduction in odds ratios from univariate to multivariable analysis, it is conceivable that observed associations with gender may be confounded by route of infection or risk group. In addition, the disappearance of anti-HCV which occurs slowly after resolution of viremia, could lead to differential underestimation of “ever infected” status in cross-sectional studies.

Another novel result of the study is that donors with higher BMI were less likely to have anti-HCV, and among antibody positives, less likely to test positive for HCV RNA. These findings were somewhat counterintuitive, as the literature implies that chronic infection with HCV may induce a “metabolic syndrome” including insulin intolerance and hepatic steatosis[31]. Similarly, HCV patients who are also obese tend to have a higher risk of progressing to high-grade fibrosis[32], perhaps due to increased inflammatory markers observed in patients who are both HCV positive and obese[33, 34]. Obesity is also a risk factor for hepatocellular carcinoma both with and without HCV infection[35]. On the other hand, a study which specifically excluded obese subjects and those with hepatic steatosis found no effect of HCV on insulin resistance[36]. Perhaps most similar to our own results, a recent large population-based study in Taiwan found that anti-HCV positive subjects tended to have lower levels of both cholesterol and triglycerides[37]. Perhaps the key to reconciling these conflicting observations is to separate studies of generally healthy populations such as blood donors, where obesity may have a protective effect against HCV infection, from studies of patients with liver disease, where obesity and HCV may represent co-morbidities. One study suggested an underlying genetic explanation for these observations with its finding that LDL receptor polymorphisms may modulate immune response to HCV[38]. Alternatively, an inflammatory state including macrophage infiltration of fat pads and increased production of IL-6 and TNF∀ has recently been linked to obesity and could contribute to the resolution of HCV infection[39, 40].

Finally, significantly higher prevalence of HCV clearance at one blood center was a puzzling result. One hypothesis is that this was due to differences in HCV genotype. Higher prevalences of the antibody positive/RNA negative state have been reported from countries with genotypes other than type 1[41, 42]. On the other hand younger injection drug users in San Francisco had a higher ratio of HCV genotypes 3 to 1 than older IDU, but their incidence of HCV clearance did not differ by genotype[28]. Blood donors in the Southern U.S. had the highest incidence of new, antibody negative/RNA positive HCV infections but did not show more non-1 HCV genotypes[43]. This suggests another possible explanation for higher proportions of HCV RNA negative status in Georgia. Most HCV RNA clearance occurs within the first year of infection[28] and may be followed by disappearance of anti-HCV over several years[44]. Therefore, a higher incidence of new HCV infections at our Georgia center during the time frame of the study could lead to oversampling of resolving infections that had lost RNA but not yet anti-HCV compared to a lower incidence center where most infections would be older, chronically RNA+.

Strengths of the current study include its large size and representation of generally healthy blood donors from several blood centers across the United States. All Anti-HCV and RNA testing was performed in parallel and with state-of-the-art assays under the rigorous quality control typical of blood center laboratories. Weaknesses include the cross-sectional design which does not allow true determination of cause and effect, and the possibility of residual confounding of reported associations by measured and unmeasured variables. Although blood center RNA performed on individual samples is generally more sensitive than assays used in the clinic[45], there may be some misclassification of RNA status due to testing in pools of 16 or 24 instead of undiluted specimens, although this effect has been shown to be small[46, 47]. Finally, although we believe blood donors are more comparable to the general population than are hospital or clinic patients with HCV infection, conclusions derived from blood donors may not be directly comparable to the general population or to patients because of selection for better health.

In conclusion, this study provides additional evidence that HCV prevalence has declined substantially among blood donors since 1993, consistent with culling of seropositives and a birth cohort effect. We found a novel association between HCV prevalence and gravidity that will be important to replicate in other U.S. studies because of the public health implications regarding potential iatrogenic transmission in the obstetric setting. Finally, our intriguing results regarding an inverse association of obesity with both HCV infection and viremia may influence the direction of future research on the genetics and metabolic effects of chronic HCV infection, particularly among more healthy HCV carriers as opposed to those with pre-existing liver disease.

Appendix

The Retrovirus Epidemiology Donor Study - II (REDS-II Study Group) is the responsibility of the following persons:

Blood Centers

American Red Cross Blood Services, New England Region: R. Cable, J. Rios, R. Benjamin

American Red Cross Blood Services, Southern Region/Department of Pathology and Laboratory Medicine, Emory University School of Medicine: C.D. Hillyer, K.L. Hillyer, J.D. Roback

Hoxworth Blood Center, University of Cincinnati Academic Health Center: R.A. Sacher, S.L. Wilkinson, P.M. Carey

Blood Centers of the Pacific: E.L. Murphy (University of California San Francisco and Blood Systems Research Institute), M.P. Busch and B. Custer (Blood Systems Research Institute), and N. Hirschler (Blood Centers of the Pacific).

The Institute for Transfusion Medicine: D. Triulzi, R. Kakaiya, J. Kiss Blood Center of Wisconsin: J. Gottschall, A. Mast

Coordinating Center

Westat, Inc.: G.B. Schreiber, M. King

Central Laboratory

Blood Systems Research Institute: M.P. Busch, P. Norris

National Heart, Lung, and Blood Institute, NIH

G.J. Nemo, T. Mondoro