Department of Obstetrics and Gynecology and Microbiology, Immunology and Molecular Genetics, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA
Cousins Center for Psychoneuroimmunology, Semel Institute for Neuroscience, Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, CA
Human T-lymphotropic virus type I (HTLV-I) is associated with adult T-cell lymphoma (ATL) and HTLV-I-associated myelopathy/tropical spastic paraparesis (HAM/TSP).1 HTLV-I infection is endemic in southern Japan, the Caribbean and portions of South America, Africa, the Middle East and Melanesia.1, 2 The risk of ATL and HAM/TSP among individuals infected with HTLV-I (“carriers”) varies markedly across Japanese and Caribbean populations. In Japanese carriers, the annual incidence of ATL is comparatively high, peaks around age 60 and is approximately 3 times greater in men than in women.3, 4 The incidence of ATL among Caribbean carriers is less than one-third the rate in Japan, peaks in the forties and does not vary by gender. In contrast, the incidence of HAM/TSP is relatively high among Caribbean carriers, whereas in Japan HAM/TSP occurs at one-tenth the rate in the Caribbean.2, 3 HAM/TSP demonstrates a female predominance in both populations.
Genetic variation of the virus does not explain the population differences in HTLV-I-related morbidity.4, 5 Instead, factors such as age at infection, route and dose of infection, nutritional status, social environment, co-morbidity and/or host genetics likely contribute to the geographic differences in natural history,6 in part by influencing the initial host immune response to HTLV-I. For example, ATL appears to develop primarily in persons who acquire HTLV-I perinatally, whereas HAM/TSP often occurs in persons infected later in life.2
Efficient cellular immunity,7 including an antigen-specific (i.e., virus-specific) cytotoxic T-lymphocyte (CTL) and a type 1 cytokine response, is important for host control of most viral infections. Several serologic markers of HTLV-1 pathogenesis (i.e. “viral markers”) serve as indicators of the host immune response to HTLV-1. One viral marker is provirus load,2, 8, 9 which represents the proportion of host peripheral blood mononuclear cells with integrated HTLV-I genome (i.e., provirus). Antibodies against HTLV-I structural proteins (anti-HTLV-I), which are found in all HTLV-I carriers, serve as another viral marker and are considered indicative of viral protein expression and of the division of provirus-containing T-lymphocytes.2, 3, 8 Antibody to Tax (anti-Tax), a product of the HTLV-I tax regulatory gene that can induce viral and host gene transcription,2 is a third viral marker. Anti-Tax detection is associated with an active CTL response to HTLV-I; the absence of anti-Tax is considered to indicate a lack of Tax expression.2 Of interest, anti-Tax is observed in HAM/TSP but not frequently in ATL patients.10–12
ATL patients are generally considered to have diminished cellular immunity1, 2; patients from Japan had Epstein-Barr virus (EBV)-specific antibody responses suggestive of diminished cellular immunity13 and increased expression of the cytokine transforming growth factor β1.14 In contrast to ATL, HAM/TSP is characterized by activated inflammatory and CTL responses to HTLV-I,1, 2 including upregulation of pro-inflammatory and type 1 cytokines.6, 14 Ultimately, differential CTL responses to HTLV-I, mediated in part by cytokines, are thought to be an important determinant of the contrasting population risks of ATL and HAM/TSP.
In a recent comparison of HTLV-I viral markers in asymptomatic carriers from Jamaica and Japan,15 we observed similar mean provirus loads in both populations (3.0 vs. 3.1 log10 copies/105 cells, respectively, p = 0.26), but significantly higher mean anti-HTLV-I titer (3.6 vs. 3.2 log10 reciprocal titer, respectively, p = 0.03) and anti-Tax seroprevalence (59% vs. 39%, respectively, p = 0.002) in Jamaican than in Japanese carriers. Those viral marker patterns suggested that Jamaican carriers may have heightened anti-HTLV-I immunity and persistent HTLV-I replication, whereas Japanese carriers have mitotic proliferation of HTLV-I-infected T-lymphocytes in the absence of strong antiviral immunity.15 Of interest, we16–18 and others13, 19 have reported immune marker patterns suggestive of subclinical immune deficiency in HTLV-I carriers in Japan. However, the immune statuses of Jamaican and Japanese carriers have not been directly compared. We conducted the present study to perform the first direct comparison of immune status across these populations. Biospecimens appropriate for a direct assessment of antiviral CTL activity were not available. Thus, we tested a panel of plasma immune markers in the HTLV-I carriers from our virus marker study15 and in HTLV-I-negative persons (“noncarriers”) from the same study populations.9, 20 We also evaluated correlations between immune markers and levels of HTLV-I viral markers among carriers within each population.15
Japanese subjects were participants in the Miyazaki Cohort Study, a prospective study conducted among adult residents of 2 rural villages in Miyazaki Prefecture between 1984 and 2000 in conjunction with government-sponsored annual health examinations.9 Jamaican individuals were participants in the Food Handlers Study,20 a series of three cross-sectional surveys conducted between 1985 and 1993 among a total of 13,920 applicants for food-handling licenses from all Jamaican parishes. From these two populations, we had previously selected 51 pairs of Japanese and Jamaican HTLV-I carriers, matched by age (±2 years), sex and sample collection year (±2 years), for a comparison of HTLV-I viral markers.15 For the present analysis, we randomly selected 1 HTLV-I noncarrier from within the same cohort for each of the 51 pairs of carriers, using the same matching criteria as for the carrier pairs. Thus, the analysis included 51 matched sets of 4 participants (N = 204) that comprised 1 carrier and 1 noncarrier from each study population. Blood samples from these subjects had been stored at −80°C in the respective repositories. The study protocol followed the human-experimentation guidelines of the US Department of Health and Human Services and received approval from the institutional review boards of each collaborating institution. Each participant provided informed consent.
HTLV-I viral markers
Laboratory methods to determine seropositivity for HTLV-I, quantify anti-HTLV-I titers, detect anti-Tax and quantify HTLV-I provirus load have been described previously.11, 15, 21
Matched sets of plasma samples were tested in the same batch for each assay, by technicians blinded to population and HTLV-I status. Plasma EBV antibodies were measured by immunofluorescence assay (IFA) techniques.22–24 Titers were reported as the highest of serial 2-fold dilutions to yield a positive IFA reading.
Commercial enzyme-linked immunosorbent assays (ELISA) were performed according to manufacturer's instructions to quantify peripheral blood levels of the following markers: soluble interleukin (IL)-2 receptor-α (sIL2R; CellFree Human sIL2R, Endogen, Woburn, MA), a molecule cleaved from the surface of activated helper T-lymphocytes25; soluble CD30 (sCD30; Human sCD30 ELISA, Bender MedSystems, Vienna, Austria), which is a tumor necrosis factor receptor superfamily protein shed from activated B- and T-lymphocytes26; C-reactive protein (CRP; Virgo CRP 150, Hemagen Diagnostics, Columbia, MD), a stable marker of chronic inflammation and correlate of circulating IL-6 levels27; and neopterin (ELItest Neopterin EIA, BRAHMS Diagnostica GMBH, Berlin, Germany), which is induced by interferon-γ during innate (nonspecific) immune responses.28 We also measured concentrations of total human immunoglobulin E (IgE), a marker of B-lymphocyte activation. Total IgE levels were determined by ELISA based on a modified Mouse IgE ELISA protocol (BD Biosciences Pharmingen, San Diego, CA), using anti-human IgE monoclonal antibodies (G7-18, G7-26, 2 μg/mL), and purified human IgE standard (31.2–500 ng/mL, Calbiochem, San Diego, CA); 1 ng/mL of the Calbiochem standard was comparable to 2 IU/mL of human IgE (NIBSC 75/502).
The detection limits of the immune marker assays were 0.25 μg/mL CRP, 6.4 U/mL sCD30, 40 pg/mL sIL2R, 2 nmol/L neopterin and 31 ng/mL total IgE. Samples with CRP and total IgE levels below the detection limit were assigned a value of half the corresponding assay's limit (i.e., 0.13 μg/mL CRP or 15.5 ng/mL total IgE).
To characterize subjects' immune marker patterns, we classified levels of antibody to EBV nuclear antigen-2 (anti-EBNA2), EBV viral capsid antigen (anti-VCA), CRP, sIL2R, sCD30, neopterin and total IgE as greater than (“higher”), or less than or equal to (“lower”), the median levels observed among HTLV-I noncarriers from the same population. Because lower anti-EBNA1 titers are associated with diminished type 1 immunity,22, 29 we grouped anti-EBNA1 titers as greater than or equal to (higher), or less than (lower) the respective median titers. We dichotomized titers of antibody to EBV early antigen (anti-EA) as seropositive (≥1:20) or seronegative (<1:20). We computed the ratio of anti-EBNA1 to anti-EBNA2 titers; a ratio of 1.0 or lower (i.e., “low EBNA1:EBNA2 ratio”), which occurs when the anti-EBNA1 titer is less than or equal to the anti-EBNA2 titer, is empirically associated with deficient control of EBV replication.22, 29
All analyses adjusted for the matching factors. Regression models included age and sample collection year as continuous variables. Linear regression models and correlation computations utilized log10-transformations of the continuous immune and/or viral marker levels. To explore underlying population differences in immune status, we computed the geometric mean level of immune markers by population in HTLV-I non-carriers using mixed-effects linear regression analyses that accounted for the matching-related correlations in the data. We assessed population differences in the covariate-adjusted means by the t-test. We used the Mantel-Haenszel test, adjusted for age (<60 vs. 60+ years), gender and sample collection year (1987–1989 vs. 1990–1993), to examine population differences in the prevalence of anti-EA seropositivity and a low EBNA1:EBNA2 ratio. Odds ratios (OR) and Wald-type 95% confidence intervals (CI) were computed by logistic regression to examine the association of HTLV-I serostatus with a given immune marker category (i.e., “higher” or “lower” levels), and (in carriers) to assess the immune markers' association with anti-Tax positivity. We constructed a separate model for each immune marker, stratified by population. We also computed population-specific Spearman partial correlation coefficients (r) to estimate pair-wise correlations between each immune marker and the reciprocal of anti-HTLV-I titer and provirus load among HTLV-I carriers.
We used SAS® statistical software (SAS Institute, Cary, NC) for the statistical analyses. All statistical tests were 2-tailed. One EBV-negative individual (Japanese) and 12 persons with nonspecific anti-EBNA1 reactivity (8 Japanese, 4 Jamaican) were omitted from EBV antibody analyses.22, 23
Age, sex and blood collection year (i.e., the matching factors) were distributed similarly by study population. The analysis included 45 (22%) persons aged 28–49, 81 (40%) aged 50–59, 53 (26%) aged 60–69 and 25 (12%) aged ≥70 years. Most matched sets (42/51; 82%) were female. We obtained 146 (72%) of the 204 plasma specimens from 1990 to 1993, and the remainder from 1987 to 1989.
Immune marker levels in persons uninfected with HTLV-I
Median EBV antibody titers and the prevalence of anti-EA seropositivity did not differ notably by population in HTLV-I noncarriers (Figs. 1a–1d). In contrast, a larger proportion of Jamaican than Japanese noncarriers had a low EBNA1:EBNA2 ratio (Fig. 1e). After adjustment for the matching factors, population differences in mean anti-EBNA1 titer and prevalence of a low EBNA1:EBNA2 ratio were statistically significant among noncarriers (Table I). HTLV-I noncarriers from Japan had lower median CRP levels (Fig. 2a) and higher median levels of sIL2R (Fig. 2b) and sCD30 (Fig. 2c) than those from Jamaica. There were no apparent differences in median neopterin (Fig. 2d) or total IgE levels (Fig. 2e) by population. After adjustment for the matching factors, the population differences in mean CRP, sIL2R and sCD30 levels were statistically significant (Table I).
Table I. Pair-Wise Comparison of Geometric Mean Level and Prevalence of Immune Markers Among 102 Persons Uninfected with HTLV-I from Japan and Jamaica
EBNA1, Epstein-Barr virus nuclear antigen type 1; GMT, geometric mean titer; EBNA2, Epstein-Barr virus nuclear antigen type 2; VCA, viral capsid antigen; EA, early antigen; CRP, C-reactive protein; sIL2R, soluble IL-2 receptor-α; sCD30, soluble CD30; IgE, immunoglobulin E.
Reciprocal of the geometric mean titer, or geometric mean level, from mixed-effects linear regression analyses adjusted for age (continuous, in years), sex and year of sample collection (continuous); p-values were obtained from the t-test and reflect covariate adjustment.
p-values were obtained from the Mantel-Haenszel test, adjusted for age (<60 vs. 60+ yr), sex and year of sample collection (1987–1989 vs. 1990–1993).
A low EBNA1:EBNA2 ratio is a ratio of anti-EBNA1 to anti-EBNA2 titers ≤ 1.0.
Values below assay detection level are assigned a value of one-half the assay detection limit and included in the estimation of the geometric mean.
Immune marker patterns associated with HTLV-I infection
Among Japanese participants, HTLV-I carriers had an increased prevalence of a low EBNA1:EBNA2 ratio compared to noncarriers (Fig. 1e). The remaining EBV antibody titers did not vary significantly by HTLV-I status among Japanese subjects (Figs. 1a–1d). With adjustment for the matching variables, Japanese HTLV-I carriers were 3 times as likely as Japanese noncarriers to have a low EBNA1:EBNA2 ratio (p = 0.06) (Table II). Japanese carriers also had higher median CRP levels (Fig. 2a) and a more than 2-fold increased odds of higher CRP levels (p = 0.04) (Table II) compared to Japanese noncarriers. Levels of sIL2R, sCD30, neopterin and total IgE did not vary markedly by HTLV-I status in Japanese subjects in crude comparisons (Figs. 2b–2e) or in multivariable regression analyses (Table II).
Table II. Association of HTLV-I Seropositivity with Immune Biomarker Levels by Study Population, with Adjustment for Age, Sex and Year of Sample Collection
Among the Jamaican participants, neither the prevalence of a low EBNA1:EBNA2 ratio (Fig. 1e), nor the concentration of CRP (Fig. 2a) varied by HTLV-I serostatus. However, both were higher in Jamaican than in Japanese subjects regardless of HTLV-I status. Compared to Jamaican noncarriers of HTLV-I, Jamaican carriers had higher median sIL2R (Fig. 2b) and sCD30 levels (Fig. 2c) and 3- and 2-fold increases in the odds of higher sIL2R (p = 0.01) and sCD30 levels (p = 0.09), respectively (Table II). However, median sIL2R levels were similar in HTLV-I carriers from Japan and Jamaica (Fig. 2b). Levels of EBV antibodies, neopterin and total IgE did not differ greatly by HTLV-I status among Jamaican subjects in crude comparisons (Figs. 1a–1d and 2d and 2e, respectively) or in multivariable regression analyses (Table II).
Correlation of HTLV-I viral markers with plasma immune markers
Among Japanese carriers, anti-HTLV-I titer was significantly correlated only with anti-EA titer (Table III). In contrast, among Jamaican carriers, anti-HTLV-I titer was significantly correlated with anti-VCA, anti-EA, sIL2R, sCD30 and neopterin levels. Provirus load was not correlated with any immune marker in Japanese carriers but was correlated with anti-EBNA2, anti-EA, sIL2R, sCD30 and neopterin levels in Jamaican carriers (Table III). Anti-Tax seropositivity was more prevalent in Japanese carriers with lower anti-EBNA1 titers, higher anti-EBNA2 titers, a low EBNA1:EBNA2 ratio or detectable anti-EA, although the associations were not statistically significant (Table III). Among Jamaican carriers, anti-Tax was more commonly detected in those who had higher sIL2R, sCD30 or neopterin levels. CRP and total IgE levels were not associated with any viral marker in either population (data not shown).
Table III. Association of HTLV-I Viral Markers with Immune Marker Levels Among Japanese and Jamaican Carriers
HTLV-I, human T-lymphotropic virus type I; sIL2R, soluble IL-2 receptor; sCD30, soluble CD30; OR, odds ratio; CI, confidence interval; EBNA1, Epstein-Barr virus nuclear antigen type 1; EBNA2, Epstein-Barr virus nuclear antigen type 2;VCA, viral capsid antigen; EA, early antigen.
Spearman partial correlations of log10-transformed HTLV-I viral markers with log10-transformed serologic immune marker levels, with mutual adjustment for the immune markers and for age, sex and year of sample collection.
From logistic regression models adjusted for age, sex and year of sample collection.
We undertook the present analysis to compare patterns of plasma markers of host immune status in Japanese and Jamaican carriers and non-carriers of HTLV-I using standardized laboratory assays. Some cytokines may be unstable, or present at levels below the lower detectable limits of commercial assays, in the plasma specimens available for our investigation.30, 31 Thus, we measured soluble protein correlates of cytokine levels that are well accepted as plasma immune markers, including several that we previously measured successfully.32 We included EBV antibody profiles as markers of cellular immunity; studies have shown that the combination of lower anti-EBNA1 and higher anti-EBNA2 titers suggest poor control of EBV infection in the context of a weak CTL response.22, 29 We also measured concentrations of plasma markers of T-cell (sIL2R, sCD30), B-cell (total IgE), and inflammation or nonspecific (CRP, neopterin) immune activation.7, 25–28, 33 Compared to cytokines, the surrogate markers are considered to be more stable, more frequently detectable and measurable with greater precision in peripheral blood specimens from asymptomatic individuals.30, 31
To characterize underlying population differences in host immunity, we examined immune marker patterns in Japanese and Jamaican noncarriers of HTLV-I. In the Jamaican noncarriers we observed marker patterns that suggest less T-lymphocyte activation (i.e., lower mean anti-EBNA1, sIL2R and sCD30 levels),25, 26 and diminished CTL control of EBV (i.e., a greater prevalence of a low EBNA1:EBNA2 ratio),22, 29 compared to Japanese noncarriers. This lower level of T-cell activation may confer a lower risk of T-cell transformation and thus contribute to the relatively low risk of ATL in Jamaica. In contrast, higher mean CRP levels in Jamaican noncarriers compared to Japanese noncarriers are consistent with increased levels of inflammation or nonspecific immune activation in the Jamaicans.27, 28 The Jamaican subjects' mean CRP levels were lower than those typically associated with inflammation-related disease34; however, such subclinical increases in inflammation may contribute to the greater risk of HAM/TSP in Jamaica. The small sample sizes in the present analysis, as well as the use of surrogate rather than direct cytokine markers, warrant cautious interpretation. Nonetheless, the contrasting immune marker patterns in Japanese and Jamaican noncarriers of HTLV-I may reflect underlying population differences in host immune status. These differences may be due, in part, to genetics, nutritional status, social environment and/or co-morbidity such as parasitic burden.
We also observed marked population differences in HTLV-I-associated immune marker patterns. Diminished cellular immunity, as characterized by a low EBNA1:EBNA2 ratio,22, 29 was more frequent in Japanese carriers than noncarriers, consistent with previous reports.13, 16–19 In contrast, the prevalence of a low EBNA1:EBNA2 ratio did not differ by HTLV-I status in the Jamaican subjects, but was greater in both the carriers and noncarriers from Jamaica than in the Japanese carriers. Thus, the apparent population differences in HTLV-I-related EBV antibody patterns reflect the population differences we observed among the HTLV-I non-carriers.
Similarly, among Japanese subjects, HTLV-I carriers had higher CRP levels than noncarriers, but the median CRP levels were even higher in Jamaican subjects regardless of HTLV-I status. The HTLV-I-related increase in CRP levels in the Japanese subjects may reflect increased IL-6 expression,35 possibly induced by Tax,36 whereas the higher CRP levels in Jamaican noncarriers may have obscured any Tax-induced effects on CRP or IL-6 in the Jamaican carriers.
HTLV-I infection of T-cells results in upregulation of activation markers, including IL2R and CD30,25, 37 on infected cells. In addition, elevated plasma sIL2R38, 39 and sCD30 levels40, 41 have been reported in ATL patients and likely reflect tumor burden related to IL2R and CD30 expression by ATL cells.25, 26 HAM/TSP patients and asymptomatic HTLV-I carriers were also reported to have significantly higher sIL2R levels than noncarriers in a predominantly Caribbean population.42 We previously reported similar sCD30 levels in asymptomatic Japanese carriers and noncarriers,16 but the levels have not been well studied in Jamaican carriers. In the present analysis, we did not observe differences in sIL2R or sCD30 levels by HTLV-I status in Japanese subjects, whereas levels of both markers were higher in carriers than noncarriers from Jamaica, especially sIL2R. Those apparent HTLV-I-associated differences in sIL2R levels in the Jamaicans were due to lower levels in the Jamaican noncarriers rather than to unusually high levels in the Jamaican carriers. Therefore, the HTLV-I-related increases in these markers among the Jamaicans appear to reflect the lower background population levels of T-cell activation that we observed in the Jamaican noncarriers.
One Jamaican study has previously examined circulating neopterin levels in asymptomatic adults by HTLV-I serostatus43 and found that the levels did not vary by HTLV-I status, consistent with the present findings. We also did not observe significant differences in total IgE levels by HTLV-I serostatus in either population, in contrast to prior reports of lower total IgE levels in HTLV-I carriers than noncarriers.16, 44, 45
In analyses restricted to carriers of HTLV-I, the immune markers generally showed more correlation with viral markers in Jamaican than in Japanese carriers. In the Japanese carriers, EBV antibody patterns consistent with diminished cellular immunity had suggestive associations with anti-Tax seropositivity. We conducted a cross-sectional analysis and therefore cannot determine the temporal relation of the viral and immune marker patterns. However, the observed correlations in Japanese carriers may reflect an increase in viral protein expression and division of HTLV-I-infected T-lymphocytes in persons with diminished type 1 immunity.8 In Jamaican carriers, plasma markers of T-cell (sIL2R, sCD30)25, 26 and nonspecific (neopterin)28 immune activation were correlated with HTLV-I provirus load and anti-HTLV-I titer and had suggestive associations with anti-Tax seropositivity. Those correlations are consistent with the prediction that active host immune responses to HTLV-I in the context of persistent HTLV-I replication characterize the Jamaican carriers in this study population.15
A synthesis of the immune and viral marker data suggests several hypotheses regarding population differences in host immune status and control of HTLV-I. Among Japanese participants, noncarriers of HTLV-I appeared to have increased T-cell activation, but HTLV-I was not associated with a further increase in T-cell activation marker levels. Concurrently, the HTLV-I viral marker patterns in Japanese carriers did not indicate persistent HTLV-I replication or CTL responses to HTLV-I.15 Thus, the Japanese carriers may not have effective virus-specific immune responses despite the evidence of T-cell activation. This suggestion is consistent with the present and previous observations of diminished type 1 immunity in Japanese carriers.13, 16–19 Furthermore, in the absence of evidence for persistent HTLV-I replication, the high provirus loads we observed in Japanese carriers are most likely a result of mitotic division of HTLV-I-infected T-lymphocytes.15 The combination of ongoing but ineffective T-cell activation and mitotic proliferation of HTLV-I-infected T-lymphocytes could contribute to the increased risk of ATL that is characteristic of Japanese carriers. In contrast, in Jamaican carriers and noncarriers of HTLV-I we observed plasma marker levels that imply chronic inflammation. In addition, Jamaican carriers had plasma T-cell activation marker levels similar to those of Japanese subjects. HTLV-I viral marker levels in Jamaican carriers further indicated persistent HTLV-I replication and virus-specific CTL responses. The combination of persistent HTLV-I replication, CTL responses and inflammation is characteristic of a population at greater risk for HAM/TSP.11, 12, 15 The lysis of HTLV-I-infected T-cells by a persistent CTL response may also contribute to the comparatively low risk of ATL observed in Jamaica.
The matched study design, direct comparisons of immune marker profiles across populations and standardized laboratory assays represent unique strengths of the present analysis. Nonetheless, limitations of this study should be noted. The immune marker findings may not be directly applicable to ATL or HAM/TSP patients, as all of the HTLV-I-infected participants were asymptomatic. Also, we could not adjust for factors that may modulate the host immune response to HTLV-I, including age at infection, route and dose of infection, nutritional status, social environment, co-morbidity and host genetics. The statistical power to detect significant small associations was limited due to the small sample size. However, we included as many participants as could be appropriately matched across populations; we are not aware of ongoing studies that could yield a larger sample size for a similar analysis.
In conclusion, we observed several intriguing new findings related to population differences in host immune status in carriers and noncarriers of HTLV-I. The findings corroborate the view that host immune dysregulation, perhaps even at a modest, subclinical level, contributes to the well established differences in natural history of HTLV-I infection in Japanese and Jamaican carriers. It remains unclear why HTLV-I infection results in contrasting types of immune dysregulation in the two populations. Possible mechanism(s) may be elucidated by longitudinal studies that include ATL and HAM/TSP patients and information on potential cofactors that was not available for the present analysis. The further development of informative, easily measured markers of immune status will also aid future studies. Those studies will provide important information to characterize the asymptomatic carriers of HTLV-I who are at an increased risk of ATL or HAM/TSP.
We thank Dr. Annika Linde for technical and intellectual expertise on the EBV serology; Dr. Chung-Cheng Hsieh for statistical consultation; Mr. Larry Magpantay for the CRP, sCD30 and total IgE assays; Dr. Najib Aziz and the Clinical Immunology Laboratory at UCLA for the sIL2R and NPT assays; Dr. Hongchuan Li for quantification of HTLV-I provirus load; Dr. Takashi Sawada for detection of anti-Tax; and Dr. Francis Yellin and Dr. Christina Raker for database administration. We are also grateful to the members of the Miyazaki Cohort Study and Food Handlers Study populations for their participation in this research.