Prevalence of chronic chikungunya and associated risks factors in the French West Indies (La Martinique): A prospective cohort study

Background The chikungunya virus (CHIKV) is a re-emerging alphavirus that can cause chronic and potentially incapacitating rheumatic musculoskeletal disorders known as chronic chikungunya arthritis (CCA). We conducted a prospective cohort study of CHIKV-infected subjects during the 2013 chikungunya outbreak in Martinique. The aim of this study was to assess the prevalence of CCA at 12 months and to search for acute phase factors significantly associated with chronicity. Methodology/Principal findings A total of 193 patients who tested positive for CHIKV RNA via qRT-PCR underwent clinical investigations in the acute phase (<21 days), and then 3, 6, and 12 months after inclusion. The Asian lineage was identified as the circulating genotype. A total of 167 participants were classified as either with or without CCA, and were analyzed using logistic regression models. The overall prevalence of CCA at 12 months was 52.1% (95%CI: 44.5–59.7). In univariate analysis, age (RD 9.62, 95% CI, 4.87;14.38, p<0.0001), female sex (RD 15.5, 95% CI, 1.03;30.0, p = 0.04), headache (RD 15.42, 95% CI, 0.65;30.18 p = 0.04), vertigo (RD 15.33, 95% CI, 1.47;29.19, p = 0.03), vomiting (RD 12.89, 95% CI, 1.54;24.24, p = 0.03), dyspnea (RD 13.53, 95% CI, 0.73;26.33, p = 0.04), intravenous rehydration (RD -16.12, 95% CI, -31.58; -0.66 p = 0.04) and urea (RD 0.66, 95% CI, 0.12;1.20, p = 0.02) were significantly associated with the development of CCA. For the subpopulation with data on joint involvement in the acute phase, the risk factors significantly associated with CCA were at least one 1 enthesitis (RD 16.7, 95%CI, 2.8; 30.7, p = 0.02) and at least one tenosynovitis (RD 16.8, 95% CI, 1.4–32.2, p = 0.04). Conclusions This cohort study conducted in Martinique confirms that CCA is a common complication of acute chikungunya disease. Our analysis emphasized the importance of age and female sex for CCA occurrence, and highlighted the aggravating role of dehydration during the acute phase. Early and adequate hydration were found to reduce the risk chronic chikungunya disorders. Trial registration clinicaltrials.gov (NCT01099852).

Chikungunya, a Makonde term meaning "that which bends up" [23], is an acute illness similar to dengue that is characterized by the abrupt onset of high-grade fever, followed by constitutional symptoms such as polyarthritis, musculoskeletal pain, headache, and skin involvement [24], and in rare cases by severe manifestations such as encephalopathy [21,24], acute hepatitis [25,26], myocarditis [27][28][29], and multi-organ failure. The acute phase (viremic period) lasts from 5 to 10 days on average, and is resolved within a few weeks. In May 2015, the World Health Organization [30] defined a person with chronic chikungunya as a "person with previous clinical diagnosis of chikungunya after 12 weeks of the onset of the symptoms presenting with at least 1 of the following articular manifestations: pain, rigidity, or edema, continuously or recurrently." Chronic arthralgia (i.e., "suspected chronic chikungunya arthritis" or "CCA") is a common complication post CHIKV-infection. Some studies [31][32][33][34][35][36][37] have reported that 40 to 60% of patients with acute chikungunya experience significantly impaired quality of life in the long term. Other studies have found that patients affected by CCA were significantly older [31,33,38,39], were more likely to be female [6,31,39,40], had severe initial rheumatic symptoms [39,41,42], and had high CHIK-specific IgG titers [38,41]. Martinique is a French overseas department with a population of nearly 400,000 and located in the French West Indies. The first autochthonous cases of chikungunya were described in the region in November 2013 (in December 2013 in Martinique) [43]. The Asian lineage was identified as the circulating genotype [44]. At the end of the outbreak (in approximately January 2015), the number of affected people was estimated at 145,000 (36% of the population) [45]. The aim of the present study was: first, to determine the prevalence of CCA at 12 months of follow-up, and second, to search for acute phase factors significantly associated with chronicity.

Ethics statement
Ethical clearance was obtained from the French National Agency for the Safety of Medicines and Health Products (ANSM) (n˚IDRCB 2010-A00282-37) and from the French Committee for the Protection of Individuals. Written and signed informed consent was obtained from all participants.

CARBO presentation
In June 2010, the Department of Infectious Diseases at the Hospital of Fort de France initiated a descriptive and prognostic study of dengue in the French West Indies and French Guiana known as the "DAG" Study. This study was based on a hospital cohort of children and adults with suspected dengue and had as its main objective to identify the predictive factors of severe dengue. Following the introduction of CHIKV in Martinique, the inclusion criteria of the DAG study were extended to chikungunya cases, and the protocol was adjusted accordingly (namely by including a broad description of rheumatic disorders specific to chikungunya disease). In May 2014, the "DAG" study became the "DAG-2" study, and included a cohort composed of 2 groups: the dengue group and the chikungunya group. This study is registered on clinicaltrials.gov (NCT01099852). The DAG-2 study is still ongoing (the duration of follow-up is 36 months), and the present analysis was conducted at 12 months of follow-up.
At the beginning of 2016, the DAG-2 cohort was extended to other arboviruses (including the Zika virus), and was renamed "the CARBO cohort." This study was conducted at Pierre Zobda-Quitman Hospital in Fort de France, capital of Martinique, in the departments most likely to receive patients infected with CHIKV (the emergency department and the department of infectious diseases). As a result, the cohort of patients was mainly ambulatory and reflected the health care-seeking population affected by the epidemic in Martinique.

Study population
Patients were included in the cohort between December 19, 2013, and December 4, 2014. Inclusion criteria were: age � 16 years of age, ability to sign informed consent, blood sample positive for CHIKV RNA via qRT-PCR, and onset of symptoms (headache, arthralgia, myalgia, fever, rash, fatigue) �7 days.

Data collection and follow-up
The DAG questionnaire was adapted to the objectives of the study. Data were collected at the initial visit (in the acute phase), and if possible on day 3, between days 5 and 7, and between days 8 and 10 after the onset of symptoms. The following data were recorded: sociodemographic data, comorbidities and clinical characteristics at baseline (headache, arthralgia, myalgia, fever, rash, fatigue, etc.), impact on quality of life as measured by the EuroQol 5 Dimensions (EQ5D) scale (S1 Fig), and treatments used. At the beginning of the epidemic, the medical team of the infectious disease department was trained by a group of rheumatologists, one of whom was a member of the CHIKV patient care team. Patients with CCA were managed in collaboration with the rheumatology department. For patients enrolled from July 2014 onwards (i.e., after adjustment of the protocol for the DAG-2 cohort), an accurate description of affected joints (location, swelling, stiffness, arthritis, enthesitis, joint pain, etc.), and an assessment of peripheral neuropathy with the douleur neuropathique 4 (DN4, neuropathic pain 4) scale (S2 Fig) were also performed.
At 3 months (M3), patients were examined by a team physician to identify CCA cases. At 6 months (M6) and 12 months (M12), patients were interviewed by phone using the DAG questionnaire (which is made up of closed questions) to monitor persistent symptoms (in particular joint pain), the impact on quality of life (as measured by the "EQ5D" scale), and treatments received. Patients were definitively classified as having CCA if they answered NO at any time of follow-up from M3 onwards to the following question: "Do you feel completely recovered from CHIKV-infection, in other words, are you now free of joint pain, rigidity, or edema related to this infection?" A clinical examination to describe and manage remaining disorders was proposed to these patients. Only patients who felt completely recovered from chikungunya at all visits from M3 onwards were categorized as NO CCA (without chronic chikungunya disease). The global clinical assessment of CCA subjects was conducted using the "Multi-Dimensional Health Assessment Questionnaire" (MDHAQ). Patients with chikungunya are known to experience a fluctuation of their symptoms (continuously or recurrently) over time. Patients with relapsing CCA were defined as having "disorders present at 1 and/or 2 time points without recovery" and patients with lingering CCA were defined as having "disorders present at all 3 time points: M3, M6, and M12." The 12-month period of follow-up allowed for the comprehensive description of chronic chikungunya disorders and for patients to characterize the time point with the heaviest burden (M3, M6, or M12).

Laboratory tests
Only CHIKV-viremic patients were recruited in the study. Blood cell counts and biochemical tests (C-reactive protein (CRP), ionogram, urea, creatinine and hepatic transaminases) were performed upon a blood sample testing positive for CHIKV RNA via qRT-PCR (RealStar Chikungunya RT-PCR kit 1.0-Altona Diagnostics, Hamburg, Germany). A biological bank was created but never used, and the viral load of patients was not collected.

Data analyses and statistical tests
Results were expressed as mean and standard deviation or frequency (percentage), as appropriate. Comparisons between the no CCA group and the CCA group were performed using chi-square test, Fisher's exact test, or Wilcoxon test, as appropriate. Results were reported as risk difference with 95% confidence interval.
A predictive set of factors associated with CCA were selected using the least absolute shrinkage and selection operator (LASSO) method, after multiple imputations with chained equations to address missing data for variable selection. The penalty term required by the LASSO method was selected using 5-fold cross-validation. Factors with less than 10% of missing data were considered for the analysis. Factors selected by LASSO have non-null coefficient. A final non-regularized logistic regression was conducted on factors with non-null coefficient, in which the effect of each factor was reported as an adjusted odds-ratio with 95% confidence interval.
All tests were two-sided, and P values of less than 0.05 were considered significant. All statistical analyses were performed with SAS 9.4 (SAS Institute, Cary, NC, USA).

Results
A total of 247 patients were eligible and consented to participate, 193 of whom were enrolled in the study (54 patients were excluded: 51 tested negative for CHIKV RNA via qRT-PCR, 2 were not tested for CHIKV RNA due to a temporary technical problem in the laboratory, and 1 was first evaluated more than 7 days after the onset of symptoms). The flowchart of the study is displayed in Fig 1.

Overall prevalence of CCA at 12 months
Of the 193 patients, 21 were lost to follow-up (i.e., they did not attend any of the follow-up visits from M3 onwards), 3 of whom had died before M3 (causes of death were: adenocarcinoma of the prostate, multiple organ failure in the acute phase of CHIKV-infection, and unknown cause). Between M6 and M12, 1 patient died of myeloma; however, this patient was not excluded from the study because he had been reported as CCA at M3 (i.e., prior to his death). Five subjects had an undetermined CHIKC status at all follow-up visits, and were therefore excluded from the analysis.
A total of 167 patients were classified as either CCA or no CCA. Close to half the patients missed the M3 visit, as only 46 CCA patients and 38 no CCA patients were counted at the time. However, a greater number of patients were evaluated at M6 and M12, respectively 142 (85%; 51 CCA and 91 no CCA) and 143 (86%; 45 CCA and 98 no CCA) patients. Of the 167 participants, 87 had experienced CCA by M12, giving an overall prevalence of 52.10% (CI 95% 44.5-59.7). Ten CCA patients attended only 1 visit, and so their CCA profile ("relapsing" or "lingering") could not be assessed. Of the 77 CCA patients, 64 (83.12%) had relapsing CCA and 13 (16.88%) had lingering CCA.
The clinical and biological characteristics of patients at the onset of disease are displayed in Table 1.

Identification of risk factors for chronic outcome (CCA)
Characteristics at disease onset in the 2 groups (no CCA and CCA) were compared to identify initial predictive factors for CCA. These characteristics are detailed in S1 and S2 Tables. Significant results are listed in Table 1. CCA patients were mainly women aged over 50 years (Fig 2). Overall, 58% (61/106) of the women developed CCA, versus 41% (24/59) of the men (sex data available for 165 patients). In univariate analysis (  Table 2). Comorbidities (cardiovascular or rheumatologic diseases) were not found to be a risk factor for CCA (Table 1). Impact on quality of life, as measured by the EQ5D scale in the acute phase, was not significantly different between CCA and no CCA patients (Table 1).
Of the 77 chronic patients who answered at least 2 of the 3 questionnaires, 64 (83%) had relapsing joint symptoms and 13 (17%) had lingering joint symptoms. Moreover, 54 out of 87 CCA patients were clinically evaluated in the chronic phase (after adjustment of the protocol for the DAG-2 cohort), with joint pain (87%), stiffness (33.3%), and swelling (22%) being the most common chronic symptoms (Table 3). Enthesitis (13%), arthritis (7.4%), and tenosynovitis (5.6%) were less frequently observed. Ankles (31.5%) and knees (29.6%) were the most affected joints, while the proximal interphalangeal joints of the hands (24.1%), the wrists Diagram of the study population. "Unable": patients unable to answer the question "Do you feel completely recovered from CHIKV-infection, in other words, are you now free of joint pain, rigidity, or edema related to this infection?"; "missed visit": not present at a specific time point; "lost to follow-up": patients who missed all follow-up visits from M3 onwards; "death": dead patients; "CCA": patients with chronic chikungunya arthritis; "no CCA": patients without chronic chikungunya arthritis.

Discussion
This was the first prospective cohort study of CHIKV-infected cases to be conducted in Martinique during the 2013-outbreak. The overall prevalence of chronic chikungunya in our cohort was 52.10% (CI 95% 44.50-59.70), which confirms that it is a common sequela after acute infection. Indeed, previous studies have reported prevalences ranging between 53.70% and 87.20% globally [6,[31][32][33]-except in India, where prevalences have been shown to vary between 4.10% and 7.00% [46,47] (possibly due to viral or population specificities). During previous outbreaks, age over 50 and female sex were shown to be the main risk factors for chronic chikungunya disease [6,32,34,40,41,48,49]. Some studies focusing on gender have suggested that the menstrual cycle [50] and ovulation influence the production of proinflammatory cytokines by monocytes [50,51] and that estradiol plays a role in the increase of antibody production [52,53]. However, these hypotheses have not been fully tested.
In our study, increased urea (p = 0.02) and reduced frequency of intravenous rehydration (p = 0.04) were associated with the development of CCA in univariate analysis. This finding suggests that initial dehydration is one of the potential factors leading to the development of CCA, just as it can cause complications in other arbovirus infections [54,55]. Our results also indicate that vertigo (p = 0.03) and headache (p = 0.04) are symptoms of dehydration, and that vomiting (p = 0.03) is an aggravating factor for dehydratation. Note that vomiting was a significant factor in the multivariable analysis, along with age and sex. The difficulty in diagnosing dehydration and the lack of human and material resources explain why rehydration was not systematically performed in our cohort of patients with CHIKV. Because cartilage is largely composed of water [56], dehydration may result in its degradation by causing cracks and increasing fragility [57,58]. A study by Sahni et al. [59] on the impact of prolonged dehydration on osseous tissue and calcium phosphate metabolism in mice concluded that prolonged water deficiency leads to decreased bone mass, with an increased rate of bone resorption and mineralization defect. It is also established that calcium metabolism disorders stimulate parathyroid hormone (PTH) secretion and activation of bone resorption by osteoclast. In our study, no difference in calcemia was observed between CCA and no CCA patients in the acute phase. However, the rate of calcium was available for only 61 patients (as our study sample was small), and this one-off measure does not reflect metabolic disorders over the medium to long term. In fact, hydration may help dilute pro-inflammatory cells and cytokines recruited in joints and may enhance anti-inflammatory response. While articular cartilage is a mechanically highly challenged material with very limited regenerative ability, recent data by Boettcher et al. [60] have shown that a broad range of mechanical and structural properties of cartilage can be restored after rehydration with physiological salt solution.
CHIKV has been shown to damage cartilage [61] and synovial tissue [62]. In view of this, we make the hypothesis that patients suffering from dehydration at the onset of the disease are at higher risk of subsequently developing damaged cartilage, which may lead to more severe joint injury, slower recovery, and chronic injury. While it has been suggested that the intensity of symptoms at presentation is predictive of CCA [41], our results do not allow us to formally respond to the hypothesis we have formulated: dehydration during the acute phase of CHIKV infection is a cause of CCA. Indeed, we did not report on the occurrence of dehydration per se. Future studies should explicitly test this hypothesis (i) by accurately quantifying the volume of liquid ingested by patients during the acute phase in prospective studies; (ii) by conducting a randomized clinical trial comparing high intravenous and oral rehydration versus rehydration advice with subsequent quantification. Lastly, dehydration may contribute to the increase in viremia through serum concentration, as some studies have reported higher CHIKV plasma viral loads in CCA patients [62,63]. However, we did not investigate this parameter in our study due to a lack of funding.
Some studies have described the inflammatory and autoimmune processes in the acute and chronic phases of chikungunya disease [62,71]. Several of them have identified a positive relationship at disease onset between CRP level [34,65], pro-inflammatory cytokines [62,71], and severity at presentation [41,42,64]. However, no autoimmune markers for CHIKV have been highlighted in observational studies [34]. In other studies, biomarkers associated with CCA were IL-6 secretion [63] and GM-CSF at disease onset [62,63]. In a study conducted in Reunion Island, the proper role of specific IgG titers in predicting CCA was observed and positively correlated with age, female sex, and severity of initial rheumatic symptoms [41]. Yet, this study found no association between CRP level at disease onset and CCA.
The main clinical characteristics of chronic patients in our study were consistent with previous descriptions [8,24,38]: polyarthralgia, articular stiffness, swollen joints, and arthritis. Moreover, as was the case in previous studies [38,[72][73][74], enthesitis and tenosynovitis were observed in our cohort. One study reported relapsing arthralgia for the majority of examined patients [40]. In our study, the development of CCA was associated with at least 1 tenosynovitis or 1 enthesitis in the acute phase. These 2 conditions, which are highly specific to symptomatic CHIKV infection, have not yet been identified as risk factors for CCA [75][76][77].
Some studies have highlighted the impact of CCA on daily activities and quality of life [35][36][37][38]41]. Patients suffering from CCA have been shown to require a multidisciplinary medical care program [78,79]. One study recommends long-term follow up by a general practitioner, often in cooperation with a rheumatologist or a rehabilitation physician [80]. Chikungunya disease and especially CCA with relapsing sequela have been shown to induce anxiety, stress, and even a nervous breakdown [81,82]. Medical treatments include analgesics, anti-inflammatory drugs, corticosteroids, and in severe cases anticonvulsant and immunosuppressive therapy [79,83,84]. Moreover, treatment may be combined with physiotherapy, physical therapy, and psychiatric and psychological evaluation [81,[82][83][84][85]. Long-term disabilities caused by chikungunya are costly at the individual and collective scales. The economic impact of the 2005-2006 chikungunya outbreak in Reunion Island was estimated at 51.63 million US dollars [86].
One of the strengths of our study was the early inclusion and prospective follow-up of patients located at the epicenter of the epidemic. Another strength was that patients were examined physically by a clinician investigator at disease onset and at M3 to monitor clinical manifestations. This objective assessment of symptoms probably prevented an overestimation of symptoms on the part of patients, though some data (e.g., those collected via joint pain assessment) remain subjective.
Our study also has limitations that must be acknowledged. First, the phone interviews conducted at M6 and M12 may have led to a declarative bias. The rate of lost to follow-up was high, though phone calls helped to reduce it. Second, the study cohort did not fully reflect the population affected by the epidemic because subjects were not sampled from the general population; thus, data for elders may be missing as their symptoms are atypical and easily misdiagnosed [87,88]. Third, the plasma viral load of CCA and no CCA patients could not be measured due to a lack of funding; however, future studies could exploit our biological bank to assess this variable and to conduct other immunological and virological investigations.
To conclude, this first prospective study conducted during the chikungunya outbreak in Martinique has described the course of the disease in the local population and has estimated the burden at 1 year. Our findings confirm that more than half of adult symptomatic infections become chronic, with age and female sex as major risk factors for incident chronic chikungunya. The predictors of chronicity identified in the acute phase lead us to suggest that dehydration is a compounding factor in the occurrence of persisting chikungunya-associated rheumatic disorders. This hypothesis opens up new avenues for prevention and research in CHIKV pathways.