Assessment of a Retrovirus Sequence and Other Possible Risk Factors for the Chronic Fatigue Syndrome in Adults

Methods

Participants and Study Design

Two parallel matched case–control studies were conducted: one to compare the presence of retroviral markers in blood samples and the other to investigate other potential risk factors. Case-patients were identified in the Atlanta metropolitan area through the CDC CFS surveillance system [11], in which sentinel physicians identify patients with at least 6 months of unexplained fatigue or “unwellness.” After the referred patients provide extensive clinical and psychological histories and their medical records are reviewed, they are classified into four groups by a panel of physicians on the basis of standard criteria. To be classified as a CFS case, a person must have reported at least 6 months of unexplained fatigue with 50% or greater reduction in either energy or activity level and must have met all other criteria of the published CFS case definition [12]. Between September 1989 and August 1991, 26 CFS case-patients were identified, 21 of whom agreed to participate in the studies.

For the retroviral marker study, each patient with CFS was matched with a healthy CDC employee for age (± 5 years), gender, and race (CDC control). Persons reporting a chronic fatigue-like illness or who were employed in a retrovirology laboratory were excluded from consideration. Each CDC control was paid $5 for participating in the study. For the risk-factor study, each patient with CFS was matched with two controls for age (±5 years), gender, and race using a random-digit dialing system within the patient's telephone prefix (neighborhood controls). Each neighborhood control was paid $50.

Retroviral Studies

Blood samples were drawn from patients with CFS and their matched CDC controls within 1 day of each other. Each heparinized specimen was given a numeric label and sent to the laboratory within 6 hours of venipuncture. All specimens were assayed for retroviral proviral DNA. A random subset of nine patients with CFS and nine matched controls was tested for serum antibodies against HTLV-I/II by Western blot and synthetic peptide enzyme-linked immunosorbent assay (ELISA) [13, 14].

Preparation of Genomic DNA

Peripheral blood lymphocytes were prepared from Ficoll-Hypaque gradients of heparinized blood samples and lysed with proteinase K [15]. Total leukocytes from the same blood samples were also separated by lysing twice with 3 volumes of ammonium chloride solution (0.85% NH_4Cl, 1 mM TRIS; pH 7.2) with two subsequent washes with phosphate-buffered saline. The leukocyte suspensions were lysed in a similar fashion. Cells from Mo-T (HTLV-II infected) and Hut-78 (uninfected) cell lines lysed similarly were used as positive and negative controls, respectively. The DNA was then phenol/chloroform extracted from lysate aliquots and ethanol precipitated [16]. The template competence of all DNA-lysate preparations was verified by polymerase chain reaction amplification, using primers from the myeloperoxidase gene [17].

Polymerase Chain Reaction Assays

Aliquots of 2 µg of DNA (or 50 µL of lysate) from peripheral blood lymphocytes and leukocytes of patients and controls were amplified for the HTLV-II gag gene fragment under conditions optimized for minimal sensitivity of 30 Mo-T cells (or 200-pg of Mo-T DNA): 1.25 mM Mg⁺⁺, and 40 cycles of 1 minute denaturation (94 °C), annealing (55 °C except for the first 6 cycles at 45 °C), and 1 minute extension (72 °C), with an additional ramping time of 30 seconds separating the annealing and extension steps [15, 18]. Only experiments that showed optimal signals with the sensitivity control and no signal with the negative control or the amplification cocktail control (all reagents except DNA template) were considered acceptable. An HTLV-I specificity control (MT-2, an HTLV-I-infected cell line) was included. The amplified products were subjected to electrophoresis on 1.5% agarose gels, were hybridized by the Southern blot method to a Phosphorus-32-labeled-probe, and were then exposed for 1 to 7 days [10].

Measurement of Risk Factors

To investigate potential risk factors, patients and neighborhood controls were first interviewed using a questionnaire to elicit basic demographic information, clinical data related to case definition, possible exposures to retroviruses, and other possible risk factors for CFS (Table 1). They were then asked to complete a self-administered questionnaire concerning sexual practices (Table 2). Controls were interviewed approximately 2 to 3 months after patients were interviewed.

Statistics

P values, odds ratios, and exact 95% CIs were calculated for stratified binomial data [19-21]. Ordered data were analyzed using the stratified exact Wilcoxon rank-sum test as indicated [22]. All P values are twice the one-tailed value unless otherwise specified and are calculated based on matched analyses. The exact tests for stratified binomial data and the exact Wilcoxon rank-sum test were computed using StatXact software (Metha and Patel, Cytel Software Corporation, Cambridge, Massachusetts).

Results

The 21 patients had onset of CFS between 1979 and 1990, with most (12 of 21) experiencing onset between 1986 and 1990. The median age at onset of CFS was 34 years (mean, 33 years; range, 16 to 51 years); the median interval from onset of CFS to interview for this study was 4.5 years (range, 1 to 12 years). Patients were all white, non-Hispanic, and native-born U.S. citizens; 18 were women.

Retroviral Studies

All samples from DNA and lysate preparations of both peripheral blood lymphocytes and leukocytes obtained from the 21 patients and 21 CDC controls failed to amplify with the HTLV-II gag gene-specific primers under conditions optimized for minimal sensitivity of 30 Mo-T cells (or 200-pg of Mo-T DNA). A representative autoradiograph from these studies is shown in Figure 1. The 18 serum samples (9 from patients and 9 from controls) tested for HTLV-I/II antibodies by Western blot and peptide ELISA were also negative.

Other Possible Risk Factors

Before becoming ill, the patients' median household income was $40 000 to $50 000, and half had completed at least 2 years of college. No statistically significant differences in these demographic variables were demonstrable between patients and controls at either the time of interview or at the time of illness onset. Patients were less likely to be living with children than were controls, both at the time of the interview (5 of 21 patients [24%] compared with 26 of 42 controls [62%]; P < 0.01) and before illness onset (6 of 21 patients [29%] compared with 24 of 42 controls [57%]; P = 0.01). The mean household size was smaller for patients than for controls at the time of interview (2.3 compared with 3.3 household members, respectively; Wilcoxon ranked-sum test, P = 0.002). Consistent with these data, women with CFS were more likely to be nulliparous at the time of interview than were women without CFS (9 of 18 female patients [50%] compared with 9 of 36 female controls [25%]; P = 0.05).

Risk factors for retroviral infection, including sexual behaviors, were not significantly different between case-patients and controls (see Tables 1 and 2). These factors included a history of sexually transmitted diseases, which may be considered a surrogate marker for the frequency of sexual behaviors conducive to retroviral transmission.

Patients were more likely than controls to report having a household member or a first-degree relative ill with a chronic fatiguing illness that reduced their activity by 50% or more for a period of at least 6 months (5 of 21 patients [24%] compared with 0 of 42 controls [0%]; P < 0.01). Seven such ill persons were reported by five case-patients. However, direct telephone questioning of the five living adult relatives indicated that only the illness of two could be confirmed as being associated with a 50% or greater reduction of activity for at least a 6-month period. The first reported having recovered from documented monospot-positive mononucleosis as a teenager, a condition that reduced his activity for a 6-month period; the onset of his illness was 7 years after the onset of CFS in the associated patient. The second reported an illness that began 2 years before her associated patient experienced the onset of CFS, and her subsequent clinical course was complicated by the diagnosis of hydrocephalus requiring shunt placement.

Patients tended to be more likely than controls to report a history of allergies (12 of 20 patients [60%] compared with 15 of 42 controls [36%]; P = 0.11), primarily allergies to respiratory irritants (for example, cigarette smoke, dust, and perfume). No statistically significant difference was observed in reported allergies to ingested products, such as foods (3 of 21 patients compared with 4 of 42 controls) and medications (7 of 21 patients compared with 8 of 42 controls), or in the frequency of asthma (2 of 21 patients compared with 1 of 42 controls). Patients also tended to be more likely than controls to report ingestion of nonvitamin, nonprescription health and food supplements before their illness onset (8 of 21 patients [38%] compared with 6 of 42 controls [14%]; P = 0.07). However, the reported supplements included a wide variety of organic and synthetic compounds, and no specific health or food supplement was mentioned by more than three patients. Similar proportions of patients and controls reported having drunk raw milk (4 of 20 patients [20%] compared with 8 of 42 controls [19%]; P = 1.0), which has been associated with a chronic fatiguing illness among children in Lyndonville, New York [23]. We could not confirm previous unpublished reports that patients were more likely than controls to have had a pet with a prolonged or unexplained illness (6 of 21 patients [29%] compared with 9 of 42 controls [21%]; P = 0.8).

Discussion

DeFreitas and colleagues [10] reported finding a retroviral marker and HTLV-I antibodies in 83% and 50% of adults with CFS, respectively, but in no controls. In our study, however, neither the marker nor HTLV-I antibodies could be detected in any patients with CFS who were identified through the CDC surveillance system in Atlanta nor in matched CDC employee controls under conditions in which the appropriate assay sensitivity controls were positive. Our study sample was small, but if the prevalence of this marker were 80%, the probability of finding all 21 of our patients negative would be extremely small (2 × 10^- 15). In addition, the epidemiologic data did not support an association between having CFS and having characteristics or behaviors either recognized to facilitate transmission of known retroviruses or hypothesized to facilitate the transmission of unidentified retroviruses. Taken together, these data cast doubt on an HTLV-like retroviral cause for CFS. Nevertheless, these data do not specifically exclude a novel or a known but unstudied retrovirus as a cause of this syndrome, especially if such a retrovirus has an unusual mode of transmission. Further, we would not reliably detect the described HTLV-II gag gene sequences in our patients if the frequency of this retrovirus in peripheral blood lymphocytes were below our minimum detection threshold or if such a virus were sequestered in tissues other than peripheral blood lymphocytes. Additionally, the number of patients with CFS studied was relatively small, and the diagnosis of CFS may currently be given to persons with a common clinical presentation of profound fatigue resulting from a spectrum of causes. Our results, therefore, do not exclude the possibility that retroviruses may cause CFS in some persons.

Our inability to detect the HTLV-II gag gene sequence is consistent with findings in a previous study of patients with CFS from western Scotland [24]. We have also been unable to detect the HTLV-II gag gene sequence in five patients who were arbitrarily selected from the CFS surveillance system in Atlanta who did not completely meet the standard case definition [25]. Other investigators have reported an inability to demonstrate antibody to human Spumavirus, another retrovirus, in blood from yet another sample of patients with CFS and control subjects [26]. Inadvertent contamination of specimens and other technical problems can produce false-positive laboratory results in polymerase chain reaction studies [18, 27]. In fact, the HTLV-II gag gene probe used by DeFreitas and colleagues [10] has an unusually high melting temperature (80.7 °C). In their study, both the hybridization and the subsequent washing were done at temperatures that constitute low-stringency conditions (37 °C) and may allow nonspecific binding of the probe to endogenously amplified material of the expected size, as was found by us and others [24]. Coupling this nonspecific phenomenon with un-conventionally long exposure times (5 to 7 days) and insufficient blinding could result in false-positive interpretations and may explain the previously reported association.

Much of the risk-factor study was exploratory, and therefore the three positive suggestive associations (P ≤ 0.07) we observed could represent false-positive results because of the multiple comparisons. In addition, historical data in our study must be interpreted cautiously because of the long interval between CFS onset and interview. Although women in our cohort of patients with CFS tended to be more likely than female controls to be nulliparous, insufficient information was collected to evaluate related issues that may affect parity. Our finding that patients with CFS were more likely than controls to have ingested nonprescription health supplements before their illness onset may reflect a predilection of these patients for nonstandard therapies [28]. One other study found an increased proportion of the patients with CFS reporting other family members with symptoms of CFS [23]. On the basis of the brief information obtained by telephone interview, however, most of the reported illnesses in our study would not meet the working case definition of CFS. The increase in reports of CFS-like illness among family members of patients may be due, in part, to an increased suspicion among patients with CFS for this syndrome in others.

We were unable to detect the previously described retroviral marker in a well-defined sample of adults with CFS or in controls. Epidemiologic characteristics of these patients also do not support transmission of known retroviruses as a cause for CFS. Until further supportive data emerge, physicians and other caregivers should not use this marker as a diagnostic test for CFS, nor should they indicate to their patients that convincing evidence associates CFS with an HTLV-like retrovirus.

Abbreviation

CFS: chronic fatigue syndrome