These results suggest that in San Francisco, within the ongoing HCV epidemic, the epidemiology of infecting HCV viruses has changed over time. Among younger IDU, who are also more likely to have more recent infection, non-GT 1 is increasingly prevalent. Historically, few studies have been published describing GTs in San Francisco patients. In 1999, Gish et al,  described GTs (using restriction fragment length polymorphism analysis) in 186 patients with chronic HCV (the majority presumptively IDU); 76% were type 1 (1a, 41%; 1b, 35%), 13% were type 2 (2a, 3%; 2b, 10%) and 11% were type 3a. In that sample, HCV GT was not associated with age, gender, ethnic origin, or presumptive mode of transmission. In two small groups (n = 8 each) of chronically HCV infected, active and former IDU recruited from two urban San Francisco Hospitals, 75% of active IDU (median age 47 years) and all of the former IDU (median age 49) were type 1. As patients in these studies were more heterogeneous with respect to routes of exposure, and were likely to have been infected some years before clinical assessment, these data are consistent with the suggestion that type 1 GTs were more prevalent in San Francisco IDU in the 1990's.
In Europe, Simmonds et al  showed an association between HCV GT and age, and attributed it primarily to risk exposure differences. In Germany, Schroter et al  also showed increases in GTs 3a in younger groups, including IDU and non-IDU, leading these authors to hypothesize that the introduction of new subtypes in a high risk group may rapidly impact lower risk groups. Others have shown differences in HCV GT distribution by route of exposure, with GT 3a associated with injection exposures, and GT 1b associated with acquisition through blood transfusion [21, 22, 30, 31]. In a multicentre study of HCV infection in 183 HIV-infected IDU sampled between 1984 and 2001, Van Asten et al , using phylogenetic analyses, found that subtypes 3a and 1a were most prevalent (38.3%, and 36%, respectively) and extensively dispersed throughout the countries.
The relative differences in HCV GT seen in these San Francisco patients may be due to various factors, including temporal changes in the dominant circulating virus, or potentially separate epidemics marked by differential social mixing. All three populations were sampled in San Francisco and were at high risk of HCV due to parenteral exposure, but younger patients had significantly lower odds of being infected with GT 1. African-Americans represented 62% of older IDU, only 1.9% of younger IDU, and 41.6% of HIV co-infected patients, but controlling for birth year and other factors, still had significantly higher odds of subtype 1 HCV infection compared to Caucasian patients. This differential GT distribution by racial group may be due to various factors including temporal, social (mixing), differential access to prevention, and/or possibly genetic explanations. The prevalence of IDU has decreased among younger age groups and among African-American populations in the U.S., and drug injector networks and associated risks may influence the spread of HCV GTs [34, 35] as seen with other diseases . Risk groups, in particular IDU compared to non-IDU populations, may have differential access to prevention methods by age, race or other factors that might result in differences in HCV transmission patterns or detection [37–39]. The genetic polymorphism recently associated with treatment and spontaneous clearance of HCV [14, 40] occurs with less frequency in African populations, and may account for some of the relative differences seen in HCV GTs prevalence in African Americans and possibly other racial/ethnic groups .
To our knowledge no studies have examined GT distribution patterns over several age groups (or by birth cohort) and risk group. In Spain, Serra et al (40), found that GTs varied by risk group: IDU were less likely to be infected with GT 1b and more likely to be infected with GTs 1a and 3, compared to transfusion recipients and blood donors, and GT prevalence varied over time in all groups . Our results are also consistent with others who have shown that African-Americans are more likely to be infected with HCV GT 1 [18, 19]. Significantly, a very recent report analyzing over 22,000 samples from HCV infected patients at a national reference laboratory in the U.S., found that patients infected with GT 3 were significantly younger (P<0.0001) than those with non-GT 3 (1, 2, or 4). the frequency of HCV GT 3 ranged from just 4.3% in subjects over age 70 to 20.6% in those age 21 to 30 . Our results are consistent with these new findings, and among the first to suggest that the relative prevalence of GT 3 may be increasing as younger groups are more likely to have been recently infected.
This study has some limitations that should be acknowledged. Genotyping was performed by reverse hybridization, and viruses were not sequenced; misclassification could have occurred . Patients may also be infected with multiple GTs, especially IDU who have frequent opportunities for exposure , and the assay would have only identified the dominant virus . The groups here are selected from studies of high risk populations in San Francisco, so results may not be generalizable to other risk groups. Finally, two groups (UHS and SCOPE) predominantly represent chronically infected patients, and samples were genotyped only if HCV RNA was present. If there is differential spontaneous clearance in association with HCV GT , these results would not reflect the circulating infecting GTs in acutely infected patients, but only those in chronically infected groups.