Nation-wide epidemiological study of Japanese patients with rare viral myelopathy using novel registration system (HAM-net)

Our finding that patients most commonly experience HAM/TSP onset in their 40s (Fig. 3) is consistent with the findings of previous studies conducted in Japan [15, 16]. However, we revealed for the first time that there is a long delay between disease onset and diagnosis in Japan: on average, there was a gap of 7.6 years (median 5, range 0–49, 95 % CI 6.8–8.4 years). For comparison, the gap was reported to be 5.3 years (median 3, range 0–33, 95 % CI 2.2–5.5 years) in Martinique [19] and 3.8 years (median 2, range 1–19 years) in the UK [18]. This discrepancy may be partially due to the fact that this national study includes patients diagnosed in rural areas with limited access to professionals familiar with this disease (Fig. 4). All over the world, many patients must endure a long struggle with mysterious symptoms before they are finally given a correct diagnosis, causing psychological strain and perhaps lowering their chances of managing the disease effectively [35]. Experts recommend that all spastic paraparesis patients in or near endemic areas be screened for HTLV-1 to ameliorate this problem [36, 37].

There is no standard system for monitoring the symptoms of HAM/TSP patients, and it is difficult to compare results across studies, especially internationally. One of the ultimate goals of HAM-net is to establish an effective system for physicians to track disease progress and for researchers to compare their findings. To work towards this goal, we have been gathering data using several pre-established scales and surveys as well as questions of our own design.

OMDS has been used for decades to monitor HAM/TSP patients [25], and in this study it was our primary means of measuring motor disability (Table 2, Fig. 5). We also collected similar data using the gait disturbance subsection of the IPEC-1 scale, named after the Evandro Chagas Institute of Clinical Research/Instituto de Pesquisa Clinica Evandro Chagas (IPEC) in Brazil, which was recently drafted to tally the presence and severity of HAM/TSP neurological symptoms including motor, spasticity, pain, numbness, urinary, and bowel symptoms [27]. It has been employed in a small clinical study in the UK [28]. We intend to evaluate the usefulness of IPEC-1 relative to OMDS in a future study. In this study, we reported the IPEC-1 gait sub-score and extracted data on lower limb numbness and pain shown in Table 1, finding that roughly half of all patients experienced some degree of lower limb pain and most experienced numbness.

Our questionnaire revealed that the majority of Japanese HAM/TSP patients require medical intervention to alleviate their urinary and bowel symptoms. Scores on the ICIQ-SF, I-PSS, OABSS, and N-QOL scales and surveys also indicated impaired urinary function; however, the scores were varied, reflecting differences in the nature of each questionnaire. For example, ICIQ-SF and N-QOL assess impact on quality of life, whereas I-PSS and OABSS focus on the severity of symptoms. Patients using unpleasantly aggressive means to treat their symptoms, such as catheters, may have eased their symptoms while decreasing quality of life. To the authors’ knowledge, this is the first time scores on ICIQ-SF, I-PSS, or OABSS have been reported for HAM/TSP patients. N-QOL was used as an outcome measure in a Japanese clinical trial for HAM/TSP in 2013 [38]. In future studies, we plan to use HAM-net data to compare the usefulness of these methods for monitoring urinary symptoms in HAM/TSP patients. Our findings demonstrate that although urinary and bowel symptoms are less visible, they are highly prevalent and severe in HAM/TSP patients, and it is critical that they not be overlooked by physicians during diagnosis, monitoring, and treatment.

Similar findings in South America have prompted researchers there to stress the importance of urinary complaints for the early diagnosis of HAM/TSP [20, 22]. Though they reported that the diagnosis of HAM/TSP tends to be delayed even longer for patients complaining first of urinary symptoms [22], we did not observe any difference in time from onset to diagnosis in this study (data not shown). We did, however, note that a fraction of patients experienced urinary symptoms before the onset of motor symptoms and that women were more likely than men to experience urinary dysfunction as an initial symptom. Given this data, doctors in endemic areas should be made aware that HAM/TSP may initially present as a problem with urinary function, especially in women.

HAM/TSP is well known to be two to three times more common in women than men [10], and as expected, nearly three-quarters of our subjects were female. This is thought to be because sexual transmission of HTLV-1 mainly occurs from male to female [1]; in addition, there is some evidence that the lifetime risk of developing HAM/TSP in HTLV-1-positive individuals is higher in women than men, although it is unclear why this would be the case [4].

Measuring changes in OMDS over time is an established method of estimating the rate of HAM/TSP progression. In this study, we recorded the number of years that elapsed between the onset of motor symptoms and progression to each OMDS reached so far by every patient who could recall this information (n = 360, Fig. 5). The results are consistent with the prevailing theory that while HAM/TSP is generally a slowly progressive disease, there is significant variation in progression speed, including a subset of patients who progress very rapidly. Figure 5 shows that the IQRs are quite large, illustrating this wide variation. Moreover, the medians increase steadily at first but then begin to curve around OMDS 9, presumably because only the group of patients who progress at the fastest rate ever reach the highest OMDS grades. We intend to evaluate progression speed in great detail in a future study.

Other studies have used similar techniques to evaluate the speed of motor disability progression in HAM/TSP patients; however, many of these studies use disease onset rather than onset of motor symptoms as a starting point. Using disease onset yields slightly longer time estimates because some patients experience urinary symptoms before motor symptoms begin. Therefore, for the sake of comparison, we also recorded changes in OMDS over time starting from disease onset: a median of 9 (3–15) years elapsed from disease onset to OMDS 5 (requiring unilateral support to walk), 14 (7.5–20) years to OMDS 6 (requiring bilateral support to walk), and 18 (10 – 27) years to OMDS 9 (unable to walk at all). Interestingly, Nakagawa et. al. reported in 1995 that it took only 12.4 ± 10.7 years (mean ± standard deviation) to progress from onset of disease to inability to walk [15]. Their study was large but only evaluated patients in Kagoshima prefecture, whereas the present study includes patients from 41 prefectures. Although there are no other large-scale studies from Japan on this topic, reports from overseas are also available for comparison. In the UK, Martin et al. observed that the median times from disease onset to dependency on a unilateral walking aid and subsequently a wheelchair were 11 years (95 % confidence interval: 2.8 to 17.30) and 18 years (95 % CI: 14 to 22), respectively [18]. In Martinique, Olindo et al. reported only 6 years (95 % CI: 5 to 7) from onset to unilateral aid, 13 years (95 % CI: 10 to 17) to bilateral aid, and as long as 21 years (95 % CI: 14 to 28) to wheelchair-dependency [19]. It should be noted that many of our wheelchair-bound patients are classified as well below OMDS 9 because they are still able to stand up from their wheelchairs and walk short distances with support. Thus, these values may not be directly comparable. It would benefit HAM/TSP research if there were an internationally agreed-upon standard for evaluating motor disability in HAM/TSP patients.

We assessed the impact of HAM/TSP on ADL using HAQ-DI (Fig. 6), a scale typically used for chronic rheumatic conditions such as arthritis [26]. To the authors’ knowledge, this is the first time the HAQ-DI scores of HAM/TSP patients have been reported. Since total HAQ-DI score was very well-correlated with OMDS, we inferred that this score could indeed be a useful measure for disease severity in HAM/TSP patients. Moreover, there were many cases where a difference in severity was detected using HAQ-DI but not OMDS, and vice versa, suggesting that combining these two measures could allow for more precise tracking of disease progression (Fig. 7). Importantly, OMDS was better correlated with certain HAQ-DI tasks than with others. Since tasks requiring heavy use of the lower limbs were particularly strenuous for HAM/TSP patients, it is of little surprise that those tasks were well-correlated. By contrast, the five tasks that focused on hand movements showed very little correlation with OMDS. Thus, it may be beneficial to exclude these five tasks (Q5, 6, 7, 16, 17) if in the future we create a HAM/TSP-specific ADL scale incorporating HAQ-DI elements.

We also assessed the impact of HAM/TSP on health-related quality of life using the SF-36 survey. Scores on this survey indicated that HAM/TSP severely impairs physical functioning but does not have a striking effect on other aspects of quality of life, such as social, emotional, and mental health. This may be due to the chronic, slowly progressive nature of the disease, which allows patients time to gradually adjust to their circumstances. However, there was a wide range of individual variation, and it is likely that patients in the advanced stages of HAM/TSP suffer from more quality of life deficits than the average patient. We intend to perform further analysis in future studies, determining what factors affect quality of life most and which SF-36 sub-sections are most useful for monitoring HAM/TSP patients.

HTLV-1 can be transmitted by blood transfusion, but this risk has been considered relatively negligible since 1986, when nationwide screening of blood donors for HTLV-1 was implemented in Japan [16, 39]. As illustrated by the distribution of HAM-net subjects shown in Fig. 4, HTLV-1-associated disease has spread all throughout Japan; however, our data indicates that blood donor screenings have been effectively limiting the spread via contaminated blood since 1986. In 1990, Osame et al. found that 23.8 % of HAM/TSP patients had a history of blood transfusion but that the number of patients developing transfusion-associated HAM/TSP had already dropped precipitously since 1986 [16]. For comparison, they conducted surveys among random Kagoshima prefecture residents and hospitalized neurological patients, and they found that the frequencies of blood transfusion history in these control populations were 3 and 5 %, respectively, when age- and sex-matched to the HAM/TSP patient group. If the 57 patients who received blood before 1986 are excluded from the analysis, only 5.0 % (16/323) of our patients had a history of blood transfusion. Thus, although these data are not directly comparable, it is clearly possible that some or all of these transfusions were merely coincidental. Moreover, 10 of those 16 patients who received blood post-1986 originated from HTLV-1-endemic areas in southern Japan (Kyushu and Okinawa), where the risk of sexual and vertical transmission is relatively high. Therefore, we have inferred that the 1986 measures have likely been effectively limiting the spread of HTLV-1 via blood transfusion.

Patients with a history of blood transfusion (n = 73), as compared to those without (n = 307), exhibited more severe symptoms (Table 3). This may be due to an elevated immune response following blood transfusion [40]. To the authors’ knowledge, this is the first time such a trend has been demonstrated. In fact, previous studies have reported that history of blood transfusion does not impact disease severity or progression speed [15, 17]. Those studies did however agree with our result that transfusion recipients were older than their non-recipient counterparts, which may also have contributed to their relatively poor outcomes [15, 22]. In fact, since HTLV-1 screening for blood transfusions was implemented in 1986, it could be argued that most patients who were infected by contaminated blood must be old enough to have received a transfusion at least 30 years ago. These patients might then suffer from more severe symptoms merely due to their age, not due to the mode of transmission. It may be possible to do an age-matched comparison in the future.

Our finding that 8.4 % of our subjects had a first- or second-degree relative with HAM/TSP is consistent with the findings of Osame et al., who reported that 8 % of HAM/TSP patients had a sibling, parent, or grandparent with HAM/TSP [16]. This is considered a relatively high rate of familial outbreak given that only 0.3 % of HTLV-1 carriers are estimated to develop HAM/TSP in their lifetimes [3]. We also evaluated the attributes of patients with a family history (Table 4), and our results agreed with those of previous studies, which showed that familial cases of HAM/TSP present with a younger age of onset and a slower rate of progression compared to sporadic cases [25, 41]. Importantly, this study revealed that those with a family history of HAM/TSP are also more likely to have a family history of ATLL. To the authors’ knowledge, this is the first time such a trend has been reported.

The relatively high rates of disease outbreak within families may indicate that genetic risk factors play a role in determining whether an HTLV-1 infection will develop into a serious illness. Indeed, genetic traits, namely certain HLA alleles, have been reported to confer increased or decreased susceptibility to HAM/TSP [42, 43] or ATLL [44], albeit confined to specific ethnic groups [45, 46]. We know of no genetic markers reported to affect susceptibility to both diseases.

An alternative line of reasoning involves considering that family history of HTLV-1-associated diseases implies vertical transmission of HTLV-1, i.e. mother-to-child, mostly through breastfeeding [47]. Supporting evidence for this reasoning includes that early infection would explain the comparatively younger ages of onset in the patients with familial HAM/TSP. It has been established that ATLL is linked to HTLV-1 infection in early childhood via vertical transmission, which is said to allow time for ATLL to develop after its typical latency period of multiple decades [48, 49]. Unlike ATLL, HAM/TSP is more commonly associated with HTLV-1 infection acquired during adulthood, either via sexual transmission or blood transfusion [49]; however, vertical transmission of HTLV-1 can also lead to HAM/TSP [50, 51]. Bartholomew et al. reported that 97 % of the mothers of their ATLL patients and 33 % of the mothers of their HAM/TSP patients were seropositive for HTLV-1 [51]. Thus, given the strong association between vertical transmission and ATLL, and assuming that cases of familial HAM/TSP are largely cases of vertical transmission, it logically follows that a high percentage of these patients would have a family history of ATLL in agreement with our results.

Similarly, the relatively high percentage of patients with a family history of HAM/TSP (8.4 %) compared to the rate of HAM/TSP in the general HTLV-1-infected population (0.3 %) could be explained if vertical transmission of HTLV-1 conferred greater susceptibility to HAM/TSP than sexual transmission, which is the most common transmission route [49]. Much about HAM/TSP pathogenesis is still unknown, and it is possible that the opportunity for the infection to develop over several decades could increase the likelihood of developing HAM/TSP as well as ATLL. The latency period for HAM/TSP can also be a long as multiple decades [50], which suggests that early infection would increase the likelihood of developing the disease within the carrier’s lifetime. This theory is only speculation, and there is no evidence to directly support the claim that vertical transmission increases susceptibility to HAM/TSP. However, indirect evidence abounds. Studies have shown that HTLV-1 carriers with a family history of HAM/TSP or ATLL have significantly higher proviral loads [52, 53]. Moreover, animal experiments have directly shown that vertical transmission (specifically oral transmission) of HTLV-1 results in elevated proviral load [54, 55]. Finally, it is well established that high proviral load is associated with the development of HAM/TSP [56]. Thus, it is reasonable to speculate that vertical transmission of HTLV-1 may result in young HTLV-1 carriers with comparatively high proviral loads who are more likely to develop HAM/TSP than those infected sexually during adulthood. Those who develop HAM/TSP would belong in our familial HAM/TSP patient group and likely present as younger and with milder inflammatory symptoms either due to their youth or greater tolerance for the virus afforded by vertical transmission [57]; this prediction is consistent with our results. Thus, the authors propose that a future study should be carried out to assess whether vertical transmission in infancy confers a greater susceptibility to HAM/TSP than sexual transmission in adulthood, as is the case for ATLL.

As a retrospective study based largely on patient recollections, our findings were vulnerable to recall error. Subjects may have mistakenly reported inaccurate information. Many subjects were elderly and did not have access to past medical records for reference. In future HAM-net studies, we will be able to rely more heavily on prospective analysis using data from the yearly follow-up interviews.

As with any registry, there is also the problem of recruitment bias. In order to recruit a sample representative of the general population in Japan, we sent information to all board-certified neurologists in the country, and patients were recruited from 41 of the 47 prefectures (Fig. 4). Still, a disproportionate number of patients were recruited from Kanagawa prefecture, presumably because our own HAM/TSP outpatient clinic recruits more effectively. It is also possible that many who live in rural areas did not seek medical advice for their HAM/TSP symptoms, were not diagnosed, and thus did not have the chance to learn about HAM-net, reducing our registry’s coverage. Another issue is that patients with a family history may have been more likely to consult a neurologist at the first sign of symptoms and thus more likely to join HAM-net. However, our data indicates that those with familial HAM/TSP were not diagnosed any more quickly than those without a family history (Table 4).

It is impossible to calculate the percentage of Japanese HAM/TSP patients who are enrolled in HAM-net, i.e. the coverage of the registry, because there is no official estimate of the number of HAM/TSP patients in Japan. In 1990, Osame et al. conducted a national survey and reported that there were at least 589 cases of HAM/TSP in Japan, while acknowledging that this was almost certainly a dramatic underestimate [16]. If it is true that there are 1.08 million people infected with HTLV-1 in Japan [2] with a 0.25 % lifetime incidence of HAM/TSP [3], there would necessarily be fewer than 2700 Japanese HAM/TSP patients. Considering these data as well as unofficial reports from colleagues, the authors speculate that there may be roughly 2000–2500 individuals suffering from HAM/TSP in Japan. With 383 subjects, this study would have thus enrolled about 15–20 % of the target population.

As mentioned, we plan to address certain topics in greater detail in future studies. To learn more about the variation in the rate of HAM/TSP progression, we intend to divide patients into groups based on progression speed and compare attributes between groups. We also plan to analyze which scales and surveys best reflect differences in the symptoms of HAM/TSP patients and eventually establish an effective monitoring method. In addition, we will be able to conduct prospective studies using data from yearly follow-up interviews; for example, we will be able to monitor the effects of treatments such as oral steroids over time. Finally, not only are we continuing yearly follow-up interviews for the current HAM-net subjects, but we are still accepting applications for new enrollees, and we expect the reach of HAM-net to expand over time. Since the data for this study was obtained and analyzed last year, we have already enrolled more patients, bringing our total up to 467 registered HAM-net subjects as of March 2016.