We reported the detection of the human gammaretrovirus XMRV in 67% of 101 patients with chronic fatigue syndrome (CFS) and in 3.7% of 218 healthy controls, but we did not claim that XMRV causes CFS. Here, we explain why the criticisms of Sudlow et al., Lloyd et al., and van der Meer et al. regarding the selection of patients and controls in our study are unwarranted.
Our study (1) documented the presence of a recently discovered human retrovirus, XMRV, in a high proportion of patients with chronic fatigue syndrome (CFS) in comparison with healthy controls. Sudlow et al. (2), Lloyd et al. (3), and van der Meer et al. (4) raise concerns about the cases and controls described in our study and thus the validity of our results. First, we wish to emphasize that our study was not intended to be a detailed clinical description of CFS or an epidemiological study that would relate particular symptoms, demographics, duration, pattern of onset, and the like to the presence or viral load of XMRV. The study was not, nor was it designed to be, a case-control study as Sudlow et al. (2) imply, for it was the first demonstration of the replication and production of infectious XMRV in human blood cells. The fact that a number of the patients tested were from regions of CFS outbreaks does not invalidate the clinical diagnosis. We hope that our report will stimulate the performance of many case-control studies that use appropriate virus detection. We certainly recognize that such studies will be required to determine what role XMRV plays in the pathogenesis of CFS.
Samples included in our study (1) were from CFS patients who fulfilled both the Fukuda criteria and the Canadian Consensus Criteria (CCC), regardless of severity. We regret that a sentence in the original supporting online material in (1) implied that immunological abnormalities were part of the CFS diagnosis; indeed, while many such patients do exhibit such abnormalities (5, 6), they were not required for diagnosis. All patients that met Centers for Disease Control and Prevention and CCC criteria were accepted; none were excluded. Patient samples were obtained from 2006 to 2009 and stored in the Whittemore Peterson Institute (WPI) repository. We did not state in Lisbon (7) or elsewhere that the samples analyzed in (1) were only from patients from documented outbreaks of CFS, nor did we state that the 101 patients described in (1) exhibited all the immunological abnormalities described in our Lisbon conference presentation. In fact, only 25 samples in (1) came from patients identified during the 1984 to 1988 CFS outbreak in Incline Village, Nevada. The remaining 76 samples included patients with sporadic cases from 12 U.S. states and Canada, including California, New York, North Carolina, Wisconsin, Michigan, Oregon, New Mexico, New Jersey, North Dakota, Texas, and Florida. Patients in the study were 67% female, reflecting the reported gender incidence of CFS, with an age distribution of 19 to 75 years of age (mean of 55). The healthy control population, which was similar in age and gender to the patients, was composed of healthy people who visited doctors’ offices in the western United States between 2006 and 2008. The great majority, although not all, of the patients analyzed were matched in geographic location with controls. As this was not an epidemiological case-control study, we did not attempt to discern where the patients believed they contracted CFS; at the time of sample collection, some were undoubtedly living in an area different from the location where they first became ill.
The information we provide here and in the accompanying Supporting Online Material (8) should lay to rest any concerns about “bias” or “confounding.” Again, the primary aim of the work described in (1) was not to characterize this clinical condition or to prove a cause for CFS but to demonstrate the existence of an infectious gammaretrovirus in patients who had been diagnosed with CFS. We achieved our goal using four different experimental strategies. The original description of HTLV-1 and HIV-1 involved only one or two patients (9–12), whereas we detected XMRV in 75 individuals.
We did not state that our study (1) proves the cause of CFS. A large number of infectious and noninfectious agents have been implicated in CFS, and it is that fact that makes the puzzle of CFS all the more difficult to solve. At no time have we wished to raise false hopes among a group of patients who, in general, have not been treated well by the medical research community. We are aware that many different pathogens have previously been reported to be associated with CFS but have not been proven to be causal.
We further note that no cytokine profiles were presented in (1), nor did we state that abnormal cytokine levels, altered natural killer cell activity, or particular RNase L profiles were a requirement for inclusion in the study. Unpublished comments made during a medical conference (7) exploring hypothetical connections with immune system defects, viral reactivation, and malignancies should not be used to judge the merits of the science in the published paper. Regarding the concern raised by Sudlow et al. (2) about potential “expectation bias,” we point out that the National Cancer Institute (NCI) and the Cleveland Clinic, whose scientists independently performed experiments and coauthored (1), were certainly not “established” as laboratories for the purpose of studying CFS. All samples were blinded, as mandated by the NCI and WPI institutional review board approvals. All experimental procedures were done by the same personnel, in the same physical laboratory space, under identical protocols. Investigators at NCI received 100 samples from individuals without knowing their health status; furthermore, the samples were sent to NCI directly without passing through the WPI laboratory space. Laboratory workers at the NCI and the WPI who performed the polymerase chain reaction (PCR) and immunological studies used coded, blinded samples that did not reveal the CFS status of the individuals. The WPI has examined all 218 control and 101 patient samples by both PCR and serological methods for the presence of XMRV nucleic acid and antibodies. In addition, NCI used plasma from all 100 samples they received in infection experiments with LNCaP cells. It was not feasible to examine all 101 patient and 218 control samples with all four XMRV detection methods described in (1), due to time and resource constraints.
Of the technologies used to identify and isolate XMRV in patients with CFS, PCR from DNA or cDNA from unstimulated peripheral blood mononuclear cells is the least sensitive method. We contend that the three recently published negative PCR studies (13–15) do not qualify as being studies that fail to replicate our study, as neither the same PCR methodologies were used nor did these studies draw on the additional cell culture and immunological methods that we employed to observe XMRV nucleic acids and proteins. Although we offer to send samples in which we have detected XMRV, the groups that published these results neither requested nor analyzed any samples we had found positive for XMRV in our laboratories.
Sudlow et al. erroneously state that we did not consider alternative explanations for the findings, namely that patients with poor general health may be more susceptible to viral and other infections. On the contrary, we raised as a question for future study: “Is XMRV infection a causal factor in the pathogenesis of CFS or a passenger virus in the immunosuppressed CFS patient population?” (1). We recognize that the presence of XMRV could be due to enhanced susceptibility to retroviral infection after development of CFS. A causal role of XMRV in CFS is an intriguing possibility, given the known immunosuppressive, neurotropic, and serious consequences of infection with other known retroviruses.
Supporting Online Material