Diagnostics

Acute CHIKV infection in our study was defined by either or both positive results of CHIKV-specific IgM in an IgM/IgG indirect immunofluorescence assay (IFA) (Anti-CHIKV IIFT IgM/IgG, Euroimmun, Lübeck, Germany) according to manufacturer’s instructions, and CHIKV-specific real-time reverse-transcription polymerase chain reaction (RT-PCR). Sera with positive IgG, but negative IgM result in IFA and negative RT-PCR were not considered acute infections. Convalescent sera to assess seroconversion were not available. The Landslaboratorium was the only diagnostic facility on the island that tested for CHIKV, using IFA. All sera were stored at the Landslaboratorium at -80°C, until shipment to the Institute of Tropical Medicine (ITM, Antwerp Belgium), where CHIKV-specific RT-PCR was performed as described previously. Briefly, a specific part of the nonstructural protein 1 (NSP-1) gene was amplified on the LightCycler 96 (Roche) according to the protocol described by Van den Bossche et al., with primers and probes detecting the African and Asian CHIKV strains[15,16]. Real-time RT-PCR was run for 50 cycles, and any Ct-value < 50 was considered positive. Isolation of CHIKV by inoculation onto Vero cells (ATCC^® CCL-81^™) was attempted from samples with Ct-values below 30. The E1-region of the isolates was genotyped (Sanger sequencing method) [17].

Because dengue virus (DENV) is endemic in Aruba, a DENV multiplex RT-PCR to differentiate the four serotypes (DENV1-4) was performed with an in-house test (adapted from ref. [18]) on the sera of confirmed acute CHIKV cases, to exclude co-infection.

Data collection

Clinical data from the laboratory request forms (gender, age and date of symptom onset- defined as 0 DPSO) were complemented by a structured telephonic questionnaire. The interviews were conducted by a trained infectious disease physician, 18–24 months after diagnosis, after obtaining oral consent. Patients with confirmed CHIKV aged 18 years or older, were eligible for interviewing. Participants were excluded after three failed attempts to contact them. Additional data obtained included comorbidities (diabetes, obesity, history of cardiovascular, respiratory or pre-existing rheumatological conditions, osteoporosis, osteoarthritis) and symptoms in the acute phase (fever, rash, arthralgia, myalgia, headache, respiratory or gastro-intestinal symptoms, difficulty sleeping or concentrating). When arthralgia was reported, the pattern of joint involvement was systematically scored in accordance with the 2010 American College of Rheumatology/ European League Against Rheumatism (ACR/EULAR) criteria for seronegative rheumatoid arthritis [19]. Briefly, the pattern of distribution of affected joints was scored based on the number and type of joints involved. ‘Large joints’ refers to shoulders, elbows, hips, knees, ankles and spine; ‘Small joints’ refers to the metacarpophalangeal, interphalangeal, metatarsophalangeal and wrist joints. We assigned low ACR-scores ‘1, 2 and 3’ to ‘one large joint, 2–10 large joints, and 1–3 small joints (with or without involvement of large joints), respectively. High ACR scores ‘4 and 5’ were assigned when involvement of ‘4–10 small joints (with or without involvement of large joints) or involvement of more than 10 joints (involving at least one small joint)’ were recorded. Symmetry, defined as bilateral involvement of at least one joint region, was assumed in all cases, unless participants explicitly reported involvement of a single joint [11]. Reported edema, redness and morning stiffness that lasted more than 30 minutes were recorded as clinical signs of inflammatory arthropathy. Self-reported duration of pain in adults with confirmed CHIKV was scored in six categories: shorter than 2 weeks, 2–6 weeks, 3–6 months, 6 weeks-3 months, 6–12 months or longer than 12 months. pCHIK-CPA was defined as joint pain lasting more than 6 weeks.

The frequency of medical consultations and duration of absenteeism (less than one week, 1–2 weeks, 2–4 weeks, more than 4 weeks or still not able to work) were recorded.

Outcomes

As primary outcomes, we recorded the relative contribution of RT-PCR and IgM antibody assays to the compound diagnosis of acute CHIKV infection per day of sampling post symptom onset, the duration of pain in adults with confirmed CHIKV (shorter than 6 weeks, longer than 6 weeks and longer than one year) and potential risk factors for pCHIK-CPA, lasting more than 6 weeks or more than 1 year. Secondary outcomes were the frequency of follow-up medical consultations and the duration of absenteeism.

Statistics

Continuous variables were summarized as medians and inter-quartile ranges. Categorical variables were expressed as counts and percentages. Chi-square test or Fisher’s exact test were used to test for association between categorical variables. Odds ratios (OR) with 95%-CI were calculated to identify risk factors in univariate logistic regression models. When significant at 10% level, a multivariate regression model was fitted to adjust for confounding and multiple predictors. The final model was selected using stepwise backward elimination with likelihood ratio test as the comparison test. All analyses were done in R (version 3.4.1).

Ethics statement

Ethics approval for this study was obtained from the Institutional Review Board at the Institute of Tropical Medicine and Ethics Committee of the University Hospital in Antwerp, Belgium. Approval for the collection of clinical data by telephonic survey after oral consent was obtained from the Board of the Horacio Oduber Hospital in Aruba. All subjects were 18 years or older when interviewed by a physician, who proceeded only after oral consent and documenting the obtained consent on the questionnaire form (See Supporting Information).

Diagnostics

Using IFA, CHIKV-specific IgM was detected in the sera of 210 of 269 confirmed acute cases in our cohort. Ninety-seven percent of samples were collected at 3 DPSO or later. Although plaque reduction neutralization tests to increase specificity were not performed, false positive IFA results are unlikely, because alphavirus outbreaks had not been recorded previously on Aruba, and because of the high analytical specificity of the IFA [29]. The accuracy of CHIKV diagnostic tests is obviously correlated to the time kinetics of viremia and specific antibody responses in human CHIKV infection.

CHIKV-specific IgM

The sensitivity of the IgM in the IFA only approximated 80% or higher after 6 DPSO (Fig 3). Post-hoc CHIKV-specific RT-PCR was positive in the sera of 164 of 269 (61%) cases. In the acute phase, the diagnostic yield additional to IgM detection alone was 28.1%. Other studies reported detection of IgM (and IgG) in 100% of cases from 5 or 6 DPSO onward [16][30].

Variation in epitope recognition between the epidemic strains might explain differences in sensitivity of tests that detect IgM. This was observed in an evaluation of commercial serological assays (including the Euroimmun IFA) between the two subsequent ECSA genotype CHIKV outbreaks in Singapore, one of which involved the E1-A226V mutation [31]. Variation between the Asian genotype CHIKV in Aruba and the ECSA genotype CHIKV that is used as an antigen substrate in the IFA, may have affected IgM detection in our cohort.

Another hypothesis to explain the reduced IgM detection rates in our cohort at 7 DPSO could be slower induction of CHIKV-specific antibody responses following infection with Asian-Caribbean genotype CHIKV compared to epidemic strains in previous reports because of differences in immunogenicity [32].

CHIKV-specific IgG

We detected CHIKV-specific IgG antibodies from 3 DPSO onwards. Detection of IgG this early is unexpected. Published data suggest that IgG does not appear until 6 DPSO, usually 1–2 days after IgM [13,16,30]. The time-kinetics of the CHIKV-specific IgG response in relation to IgM and RNA results seems compatible with acute infection (Fig 3). Isolated IgG-positive results (i.e. in sera with negative RT-PCR and IgM results) were a rare finding in our cohort, and occurred only at the height of the epidemic or later. We therefore believe that all positive antibody detection results indicate recent CHIKV infection during the outbreak we describe here.

CHIKV-specific RT-PCR

The highest proportion of viremic cases (88%) was found at 4 DPSO. CHIKV RNA was detected in spite of the presence of IgM and IgG in up to 45% and 30% of sera, respectively. This is an unexpected finding, for experimental observations had suggested that CHIKV-specific antibodies had strong neutralizing activity in vitro and even therapeutic potential against CHIKV infection in mice [33]. Jain et al. postulated that appearance of CHIKV-specific IgG is required to neutralize the virus [34]. In accordance, previous observations of CHIKV infection in humans coupled the development of a CHIKV-specific antibody response to a rapid decrease in viremia [13,30,31]. In studies from CHIKV outbreaks that employed both type of tests throughout the acute phase of infection, combined positive results for both RT-PCR and antibody detection based assays do not appear frequent, i.e. 1–12%. [35–38]. However, our findings are in line with an analysis at California’s National Reference Laboratory, that detected CHIKV-RNA in 34% and 36% of 376 IgM and IgG positive sera, respectively [39]. Sequencing was not performed in that study, but Asian genotype was assumed since samples were obtained from travellers returning from the Caribbean in 2014.

Interestingly, 40 to 50% of patients in our cohort were viremic beyond 7 DPSO. This contrasts with frequent reports of CHIKV-RNA declining to undetectable levels within a week [16,35,40–43]. One small study from Singapore found viremia to persist for more than a week in 30% of cases [44]. In a large cohort from India, viral RNA was detected in patients who were recruited as late as 12 DPSO [34]. In a Thai study, quantitative RT-PCR detected viremia up to 6 DPSO, but when using a second, nested RT-PCR protocol, CHIKV-RNA was found up to 9 DPSO in all of 45 patients [30]. Detection of viral RNA depends on the sensitivity of the assays employed for nucleic acid detection. Variation in CHIKV genotypes or epidemic strains might account for observed differences in viral kinetics between outbreaks.

Clinical presentation

Arthralgia, fever, myalgia, headache and skin rash were the most frequent acute phase symptoms reported by CHIKV infected patients in our Telephone survey (Table 1). Many patients complained of dizziness, but no other neurological symptoms were recorded. Sleeping and concentration disorders were recorded frequently, but it is impossible to determine whether they were caused by the infection or resulted from illness and pain. The clinical presentation in this study is largely consistent with reports from recent outbreaks, including those caused by strains derived from the Asian CHIKV lineage that affected territories in the Caribbean and South and Central America [43,45–48]. A notable difference is that as many as 20% of patients in our adult study population experienced cough, sore throat or dyspnoea. Respiratory symptoms associated with CHIKV infection have not been reported consistently, but occurred in 23–30% of cases in epidemics in the Philippines, Trinidad and Tobago, and the Dominican Republic [36,38,49]. Critical illness and death due to pneumonia and respiratory failure were reported in ECSA genotype CHIKV-infected patients during the outbreak at Réunion [50]. It remains unclear whether these cases were caused directly by CHIKV, or by co-infection with respiratory pathogens to which CHIKV infection may predispose [51].

In line with published data from the 2014 Asian lineage outbreak in the Caribbean, no CHIKV-associated mortality was recorded in our series. Mortality resulting from CHIKV infection is thought to be rare and restricted to patients with atypical presentations or comorbidities, although some authors reported case-fatality ratios of up to 5% [6,52,53].

Arthralgia

The dominant clinical symptom in confirmed CHIKV infection was joint pain. Inflammatory characteristics such as morning stiffness, edema and redness were commonly reported. As a rule, multiple joints were symmetrically affected (resulting in a high ACR-score), with the fingers, knees, ankles and feet most commonly involved. pCHIK-CPA occurred frequently.

Previous reports vary greatly in mode of assessment, description, follow-up and classification of CHIKV-associated joint pains [10]. After clinical evaluation, pCHIK-CPA has been diagnosed as arthritis, spondyloarthritis, tenosynovitis, fibromyalgia and arthralgia syndrome among others [54–57]. The pathogenesis of CHIKV-associated arthropathy remains incompletely understood. The pattern of leucocyte infiltration, cytokine production, and complement activation in CHIKV-infected joints is similar to that found in (seronegative) rheumatoid arthritis [22,58]. Clinically, inflammatory joint destruction by synovitis, erosions and joint space narrowing has been well documented [57,59].

The estimates of long-term arthralgia occurrence after CHIKV vary widely [10,60]. Studies in mice suggest that the Asian CHIKV lineage from the Caribbean induces less severe joint pathology than the ECSA genotype from Réunion [61]. Other sources of variation in disease severity between outbreaks may be differences in host response. Variation is certainly affected by differences in study design, such as cohort size, patient selection, mode and timing of assessment of joint pains, and length of follow-up. Previously, arthralgia was found to persist in 57 to 67% of CHIKV infected patients following the ECSA genotype outbreaks in Réunion and Europe [55,62–65]. These percentages, from studies with an approach and length of follow-up comparable to ours, were supported by a prospective study of viremic patients from Réunion [56]. However, well-designed studies from Sri Lanka and India involving ECSA genotype, found rheumatic pains to last considerably shorter (4.1 and 17.8% at 12 months) [66,67]. In a study of 437 patients who did not have prior joint pains from India, rheumatologists diagnosed 57% with postviral polyarthralgia, 22% with postviral inflammatory polyarthritis and 1.4% with rheumatoid arthritis, 15 months following CHIKV infection [68].

Only a few estimates of chronic sequelae of CHIKV infection in the Americas have been published. In Colombia, Rodriguez-Morales et al. found that polyarthralgia persisted during a median follow-up of 21 weeks in 44.3% of 131 infected patients [69]. In Suriname, 22.2% of 180 symptomatic and viremic CHIKV-patients in prospective follow-up, still reported arthralgia at 6 months [48].

Our data on pCHIK-CPA are consistent with these earlier data from Asian genotype CHIKV outbreaks in the Americas. They are also consistent with a meta-analysis that conservatively estimated the proportion of patients with chronic inflammatory rheumatism following CHIKV-infection at 25%, and the percentage developing arthritis at 14% [60].

Predictors of CHIKV-associated long-term arthralgia

Analysing our clinical data, three independent risk factors for arthralgia lasting longer than 6 weeks held significance in multivariate analysis: female gender, obesity and a high ACR-score. Female gender and a high ACR-score also predicted pCHIK-CPA longer than one year. Age and co-morbidities other than obesity, were not associated with long-term arthralgia. Sleep and concentration disorders were correlated with persisting joint pains, but our study design did not enable us to distinguish cause from consequence. In addition, prolonged detection of CHIKV-RNA in serum at 8 to 10 DPSO was identified as a predictor for pCHIK-CPA at 6 weeks and at 1 year.

In our study, the number of female patients was twice that of men in both confirmed CHIKV and non-CHIKV cases. We therefore postulate that the female-to-male ratio in our study reflects gender differences in health seeking behaviour, rather than genuine differences in exposure to CHIKV or frequency of symptomatic infection. The seroprevalence rates from an outbreak that involved the same CHIKV genotype Saint Martin in 2014, also showed similar rates of infection between men and women [27]. Differential susceptibility to CHIKV infection has been reported for either sex in studies from different outbreaks [70]. Female gender has been mentioned as a risk factor for long-term joint pains in many CHIKV outbreaks [48,55,71–73]. However, it has not been identified consistently, and pathophysiological evidence for a gender difference is lacking [64,74,75].

Age above 30 to 50 years was a risk factor in many reports [48,56,63,71,74–76]. In line with studies by Larrieu et al. and Win et al., that involved ECSA genotype, age did not emerge as a risk factor in our univariate analysis, which excluded participants younger than 18 years [64,73].

Obesity, a major public health problem in Aruba, predicted pCHIK-CPA. For joint pains longer than 6 weeks duration, this self-reported finding was an independent predictor of commonly associated co-morbidities such as diabetes or degenerative joint disease, which were identified as risk factors in other studies [56,63]. Obesity was also identified as an independent risk factor for CHIKV-induced arthritis in a large study from India [77]. The authors suggested that obesity may lead to a chronic inflammatory state, that is aggravated by CHIKV infection. This hypothesis merits further study, especially since others found a trend of persisting arthralgia in patients with increased body mass index, even though it did not reach statistical significance [64,75].

In contrast with previous findings, long-term arthralgia was not associated with pre-existing rheumatological disease or other comorbidities [55,56,63]. The relatively low prevalence of comorbid conditions in our cohort precludes generalization of these results.

Signs of inflammatory arthropathy and involvement of the fingers predicted long-term arthralgia in univariate, but not multivariate analysis. Instead, a high ACR-score in the acute phase was a significant risk factor for pCHIK-CPA. We therefore concur with the French guidelines for the management of chikungunya, to make an accurate assessment of the number of joints involved in the initial evaluation of CHIKV-infected patients [78].

Presence of CHIKV-RNA in serum after the first week was a strong predictor of pCHIK-CPA. This may point towards high viral loads at the time of symptom onset, which have been associated with prolonged joint pains in ECSA genotype infection [34,79]. However, another study failed to demonstrate an association between peak viral load or duration of viremia with persistent arthralgia at 6 weeks [73]. Following infection in humans, CHIKV rapidly mutates during the acute phase. Thiberville et al. observed that the increasing intra-host genetic diversity, with resulting higher viral amino-acid complexity in the acute phase, correlated with increased frequency of pCHIK-CPA at 300 DPSO [40]. The formation of intra-host quasi-species may help CHIKV to evade a mounting adaptive immune response, and could explain prolonged persistence of CHIKV-RNA in spite of detectable CHIKV-specific antibodies. The advent of affordable Next-Generation Sequencing instruments should enable us to investigate this hypothesis in the near future.

Finally, pCHIK-CPA was associated with extended periods of sick leave. Although mortality and hospitalization rates appear lower than those reported for the ECSA genotype outbreaks in Asia, the high attack rates of CHIKV in the Americas are associated with considerable morbidity. The increase in consultations and absenteeism we reported here, seem a crude measure of the direct and indirect health costs associated with CHIKV outbreaks.

Limitations

Our retrospective study was subject to a number of limitations. Recall bias could not be excluded. The time lapse between the survey and disease onset might have affected acute phase reporting in particular. The structured questionnaire used multiple choice questions, which restricted the amount of detail we recorded. Further, description of inflammatory characteristics and arthralgia classification was based on self-reporting, which was reported to overestimate the number of joints involved and the presence of signs compared to assessment by a qualified physician [62]. A prospective study has been planned to assess structural joint damage in patients with persisting pCHIK-CPA in Aruba. Public awareness of lack of specific antiviral treatment probably prevented many symptomatic patients from seeking medical consultation, and may have introduced selection bias in our cohort. As with other outbreak reports, the unique interaction between infecting CHIKV strain and host population limits generalizability of our study results.

Conclusions

We report on the first outbreak of an Asian lineage of CHIKV in Aruba. More than one fifth of cases would not have been detected if RT-PCR testing was omitted. We emphasize the need for a combined molecular and serological approach to diagnose CHIKV in a single sample during the acute phase. Rapid diagnostic tests, that meet the ASSURED criteria (affordable, sensitive, specific, user-friendly, rapid and robust, equipment-free and delivered on location) are urgently needed [80]. CHIKV-antigen detection tests are being developed, but unsatisfactory diagnostic performance across different genotypes does not permit commercialization yet [81,82].

Our survey of the clinical presentation of CHIKV infection in an adult population indicated that arthralgia, fever and skin rash were the dominant acute phase symptoms. Arthralgia was characterised by signs of inflammation in up to 71% of those affected. Forty-four percent of cases had pCHIK-CPA for longer than 6 weeks, and 26% for at least one year. In Aruba, a study has been planned to classify rheumatic patterns and to assess structural joint damage in patients with persisting pCHIK-CPA.

We identified female gender, obesity, a high ACR-score and prolonged detection of CHIKV-RNA in serum (longer than 1 week after symptom onset) as predictors for pCHIK-CPA. These risk factors should be validated in prospective studies, so they could contribute to the timely identification of CHIKV-infected patients who are likely to benefit from future therapeutic interventions, such as disease-modifying anti-rheumatic drugs or antiviral compounds [78,83–85].