Outcomes , rates and predictors of transition of isolated Raynaud ’ s phenomenon : a systematic review and meta-analysis

RESULTS: Of 856 articles captured, 7 selected studies met the inclusion criteria. A total of 4051 patients with primary Rp (pRp) and 1220 transitions to overt CTDs were recorded. The mean incidence rate of transition from pRp to a CTD was 2.65/100 person-years (standard error [SE] 1.2, 95% confidence interval [CI] 0.44–5.73). A total of 657 patients with suspected secondary Rp (ssRp) had antinuclear antibodies (ANAs) and/or capillary abnormalities; 188 transitions to CTDs were recorded, the mean incidence rate of transition from ssRp to CTD was 11.01/100 person-years (SE 4.0, 95% CI 0.11–22.12), and 135 transitions to systemic sclerosis (SSc), giving a mean incidence rate of transition from ssRp to SSc of 5.7/100 person-years (SE 2.19, 95% CI 1.02–13.19). With respect to patients with pRp, having ANAs without capillary abnormalities was associated with a risk for developing a CTD (pooled relative risks [RR] 7.63, 95% CI 2.87–20.29), whereas capillary abnormalities without ANAs resulted in a weaker risk of CTD transition (RR 5.53, 95% CI 1.45–21.06). The coexistence of ANAs and abnormal capillaroscopy significantly increased the risk of transition to CTD (RR 16.96, 95% CI 6.61–43.55).


Introduction
Raynaud's phenomenon (Rp) is a common condition characterised by episodic reversible vasospasm of the extremities on exposure to cold or emotional stress [1].It is closely related to climatic conditions [2], and its prevalence varies between studies [3,4]; for example, in the North America, the prevalence is 4 to 9% among women and 3 to 6% among men, whereas in Europe it is 2 to 21% [5][6][7].Primary Rp (pRp) is generally a benign condition characterised by functional changes in blood vessels and/or their innervation in patients without autoantibodies and normal nail-fold capillaries.Although by definition pRp should not progress to an overt connective tissue disease (CTD), some population-based studies suggest that secondary Rp (sRp) develops in 12 to 20% of subjects first diagnosed with pRp [1,8,9].Secondary Rp develops in the context of an associated disorder or condition, including many nonrheumatic conditions (e.g., the use of vibratory tools, some compounds, neoplasms), and rheumatic diseases (e.g., CTDs) [10].In relation to CTDs, Rp may be either a concomitant symptom that accompanies other more specific clinical manifestations or an early symptom of a developing CTD [1].Many studies have investigated the risk of transition from pRp to sRp and the role of different factors used during the Rp screening phase (history, examination, investigations) in predicting evolution toward CTDs among various populations [5,9,[11][12][13][14][15][16].In 1998, Spencer-Green [9] published a meta-analysis of 10 studies (639 pRp patients) that evaluated transition rates and predictors of transition to CTD.The transition rate found (3.2 transitions per 100 patient-years of follow-up) was low, especially in consideration of the time from Rp onset (1.4 per 100 patient-years).Abnormal nail-fold capillaries at study entry, even when associated with positive antinuclear antibody (ANA) tests, was the chief predictor of evolution to CTDs [9].
In recent years, significant efforts have been made to achieve an early diagnosis of CTDs and new classification criteria have been released [17,18].The present systematic review and meta-analysis aimed to add new evidence to the previous work by Spencer-Green [9], supposing that the most recent literature could provide a more accurate estimation of clinical outcomes and risk factors for transitions to sRp.

Literature search
The reporting of the study was guided by the MOOSE Statement [19].A comprehensive systematic literature search was undertaken, which included studies published between July 1996 and August 2014, and an updated search was carried out in May 2016 to capture newly published studies.The databases used were Medline (via PubMed) and Embase.The search strategy was intended to uncover all relevant papers reporting on an adult population with Rp and dealing with any aspect of transition to overt CTDs, such as incidence and risk factors.The full search strategy is provided in supplementary table S1 (appendix 1).During the first analysis, two reviewers (F.I., C.T.) independently screened titles/abstracts and the full papers were sought if relevant.Any duplicate articles were excluded.Reference lists of review articles were also examined for relevant studies.In cases of disagreement, a decision was reached by consensus or, if necessary, a third reviewer (N.U.) was consulted.

Study selection
Study selection was by N.U.and C.C. To be eligible, studies needed to meet the following criteria: adult participants (>16 years old) with Rp classified as primary or secondary; longitudinal observational studies (prospective or retrospective) and case-control studies were considered suitable in scope of review; studies reporting the incidence of transition of Rp to CTDs and/or potential risk factors; studies with a detailed description of parameters used to define and classify Rp; studies with at least the abstract in English.Exclusion criteria were: studies assessing treatment of Rp; studies involving patients with a clear definition of secondary Rp; studies not reporting the criteria for diagnosing a CTD; studies assessing occupational Rp; unpublished material, case reports, editorials, letters or reviews.Where different articles of the same author appeared to describe same patients more than once, only the article providing the most detailed information on the patients was included.

Data extraction and synthesis
Data extraction from each article was by N.U.and C.C.An independent reviewer (F.I.) ensured the quality of data extraction.Data concerning study design, country, age range, gender ratio and total number of participants, exposure to ANAs, capillaroscopy, and development of a CTD were collected in ad hoc forms.If the same study population was reported in more than one article, we used the article with data most clearly presented.Where possible, the raw outcomes were extracted (primary outcome: number of persons developing an overt CTD) for each group (exposed to ANAs and/or abnormal capillaroscopy vs not exposed) and the effect estimates were calculated.Where data retrieval from the articles was difficult, the authors were contacted by email.To assess the methodological quality of nonrandomised observational studies, we used the Newcastle-Ottawa scale (NOS) recommended by the Cochrane Collaboration.According to the NOS [20,21], a study can be awarded a maximum of one point for each numbered item within the "selection" and "outcome" categories (four and three items, respectively); a maximum of two points can be given for comparability (one item).The quality of the included studies was independently evaluated by N.U.and C.C. and rated by consensus.Studies were considered of good quality if the total score was ≥5.No studies were excluded from data analysis on the basis of the quality score grading.

Case definitions
Patients with Rp were categorised into two distinct clinical entities: (1) patients with pRp defined according to the Leroy and Medsger criteria (i.e., no history or physical findings suggestive of a secondary cause, normal capillaroscopy, negative serological findings) [22]; (2) patients with Rp associated with ANAs and/or abnormal capillaroscopy, even if associated with symptoms or physical findings suggestive of a secondary cause, but without fulfilling criteria for a definite CTD.Up to now, the latter condition has been variously defined and there is no consensus among experts.In our analysis, we refer to it as "suspected secondary Rp" (ssRp).Patients fulfilling the definition of pRp and ssRp were analysed separately for transition rates.We were aware that the ssRp group could be heterogeneous in terms of prognosis, so we decided to split the analysis on risk of transition according to the following three subgroups: (1) patients with ANAs positive without capillaroscopy abnormalities (Rp with immune signature); (2) patients with ANAs negative with capillaroscopy abnormalities (Rp with vascular signature); (3) patients with ANAs positive and abnormalities on capillaroscopy (Rp with immune and vascular signatures).As capillaroscopy abnormalities, we considered findings such as giant capillaries, decrease in the capillary number, capillary derangement and microhaemorrhages, which are required for the definition of "scleroderma pattern" in accordance with the description reported either by Maricq [23] or by Cutolo [24].

Statistical analysis
Individual data to calculate the incidence of transition were derived from the original reports from the information provided in each study.Cumulative incidences and 95% confidence intervals (CIs) were transformed into incidence rate data (incidence per 100 person-years).Relative risks (RR) and 95% CIs were calculated to present the association between risk factors (ANA positivity and capillaroscopy abnormalities) and transition to CTD.The presence of statistical heterogeneity was evaluated with the chi-squared test and a p-value <0.1 was considered significant; inconsistency across studies was quantified as the I 2 statistic, where a value >50% was considered to indicate substantial heterogeneity.A Mantel-Haenszel random-effects model for dichotomous variables was used to pool the data into a metaanalysis.
All statistical analyses were carried out using Stata V.13.0;RevMan and Comprehensive meta-analysis software were used to pool data on RR and incidence rates, respectively.

Results
In total, 856 references were retrieved in the initial search strategy in PubMed and Embase.Of these, 826 were excluded as duplicates or after title/abstract screening.Thirty articles were retrieved for full paper review, of which seven fulfilled the inclusion criteria [5,[11][12][13][14][15][16].The review flow process is outlined in fig. 1.The included studies are listed in table 1, with a description of the study design, population, setting and diagnosis at baseline [5,[11][12][13][14][15][16], as well as NOS quality ratings.Among the seven studies included, two analysed cohorts of patients with pRp, one included only patients with ssRp, and four reported data about both groups of patients.No case-control studies satisfied the inclusion criteria for this meta-analysis.
As shown in table 2A, a total of 4051 unique patients with pRp were included in six studies.Three articles [5,12,15] reported information about gender (3394 patients, 88.1% were women); five studies [5,11,12,15,16] provided data about the age at baseline (mean 43.1 years, range 39.8-47.0years); four articles [5,11,15,16] reported the average age at onset of Rp (3712 patients, mean 34.1 years, range 28.5-38.1 years) and the average of pRp duration (3712 patients, mean 8.0 years, range 5.3-11.6 years).As shown in table 2B, among five studies [5,[11][12][13][14] with 657 ssRp patients, four [5,[11][12][13] provided information about gender (281 patients, 89.0% were women) and age at the time of study entry (281 patients, median 44.1 years, range 37.5-48.3years); three articles [5,11,13] reported the average age at onset of Rp (122 patients, mean 37.0 years, range 34.5-39.0years) and in these patients Rp had been present for a mean of 5.7 years (range, 3.0-8.2years) before study entry.The effects of age and sex on transition to a CTD were not analysed owing to the lack of data.*referred to the overall population pRp: primary Raynaud's phenomenon; ssRp: suspected secondary Raynaud's phenomenon; NOS: Newcastle-Ottawa scale.A study can be awarded a maximum of one point for each numbered item within the Selection and Outcome categories (maximum four and three points, respectively).A maximum of two points can be given for Comparability.The overall score ranges from 0 to 9. Quality scale does not imply that items are of equal relevant importance.Clinical and laboratory variables were remarkably heterogeneous, including at the baseline evaluation, and the identification of precise clinical predictors of transition was limited.However, in all the seven studies (4708 unique participants), ANAs were measured and nail-fold capillaroscopy was performed.Five studies provided information about ANA assessment by use of either an indirect immunofluorescence (IFI) test on Hep-2 cells [5,12,14,15] or by both IFI and the enzyme-linked immunosorbent assay [11]; in six studies [5,[11][12][13][14]16], nail-fold capillary abnormalities were defined as proposed in the definitions of scleroderma patterns in accordance with Maricq [23] or Cutolo [24].temic vasculitis, 26 (2.1%) poly/dermatomyositis, 5 (0.4%) antiphospholipid syndrome (fig.S2).

Rate of transition from pRP to a CTD
Because of the great heterogeneity among studies regarding sample size, follow-up duration and outcomes, data on transition rates were not meta-analysed.
Because of the great heterogeneity among studies regarding sample size, follow-up duration and outcomes, data on transition rates were not meta-analysed.

Predictors of transition
All the studies identified ANAs and nail-fold capillary abnormalities as predictors of transition to a CTD.In particular (fig.2A

Statement of principal findings
The present systematic review and meta-analysis was planned to add new evidence to the previous work by Spencer-Green published in 1998 [9].According to Spencer-Green, a patient presenting only with Rp, without signs suggestive for a CTD, could be reassured: this condition is likely to have a benign course.Overall, 4708 Rp patients were included in the present analysis; according to our case definitions, 4051 had pRp and 657 had ssRp.A total of 1408 transitions to CTDs were recorded (30.1% from pRp, 28.6% from ssRp) during the follow-up.
The presence of autoantibodies (even without any specificity) should be carefully considered when assessing possible evolution to a secondary form.The coexistence of abnormalities of capillaries should raise suspicion for possible evolution to SSc and requires watchful monitoring and early intervention strategies.
Our data showed that ANA positivity is an important predictive factor for the evolution to CTDs other than SSc.
In particular, most of the transitions to CTDs were toward undifferentiated connective tissue disease and SLE; this is not surprising as Rp is one of the most frequently reported symptoms in patients with these diseases [25], although Rp is not mentioned in proposed undifferentiated connective tissue disease classification criteria, nor in the newest classification criteria for SLE [26].Therefore, these data highlight the importance of appropriate monitoring strategies for patients with Rp and ANAs.

Strengths and weakness of the study
The use of two databases might be considered a limitation.
A source of uncertainty is the great heterogeneity among studies, especially for transition rate outcome, which was probably due to great variability in clinical assessment; this could misrepresent the average rate.Our systematic literature review highlighted the variety of nomenclature for patients with Rp, positive ANAs, and/or nail-fold capillary abnormalities, even in association with symptoms or physical findings suggestive of a secondary cause but without fulfilling the diagnostic criteria for a definite CTD.This is still a matter of debate: to overcome this problem we called these patients "suspected secondary Rp" with an "immune" or a "vascular" signature.
In order to increase consistency and accuracy, we decided to consider data for pRp and ssRp separately.We are aware

Systematic review
that this choice reduced the comparability of our results with Spencer-Green analysis [9].However, we think this approach guarantees more clinically meaningful results; as confirmation, different outcomes were reported in the two conditions.Higher transition rates were reported by two retrospective studies [12,27], thus we could hypothesise that there was an assessment bias in the baseline evaluation, which artificially increased the evolution rate.
In the analysis of ssRp transition to SSc, very low heterogeneity among studies was found.This is in line with an increased awareness of SSc-specific prognostic factors for early diagnosis of the disease.All studies included in this meta-analysis derived from tertiary referral centres for CTDs; this could represent a referral bias, increasing transition rates to CTD because the most severe cases are generally addressed to tertiary hospitals.
Finally, data on evolution to non-SSc CTDs were available in only two studies [5,12]; the focus on SSc could have caused underestimation of transition to CTDs other than SSc.

Comparison with other studies
These results are in line with Spencer-Green's [9] previous observation: a transition rate to CTDs of 3.2 ± 2.0 per 100 person-years (range 0.8-7.0).
We decided to analyse separately patients with Rp and ANA positivity and/or capillary abnormalities (immune and/or vascular signature Rp) because of the growing evidence of the role of these variables on disease outcomes [1,10].As expected, the transition rate was significantly greater in the ssRp group (10.61 and 6.57 per 100 personyears for transition to any CTD and to SSc, respectively).Indeed, our analysis confirms that the two conditions (pRp and ssRp) have different prognoses and should be considered separately; besides, the presence of ANAs and capillaroscopy abnormalities (especially coexistence of both conditions) are associated with significant risk of transition to a definite CTD and, in particular, SSc.These data confirmed that patients with Rp and SSc-marker autoantibodies and/or typical nail-fold capillaroscopic findings, but no other organ manifestations (thus not fulfilling classification criteria for SSc), may be identified as "early SSc" and often present preclinical organ dysfunction [28,29].Our data are also in line with the Very Early Diagnosis of Systemic Sclerosis (VEDOSS) project in which the presence of specific findings (including Rp, puffy fingers and ANA positivity), defined as "red flags", is recommended to raise a strong suspicion of a diagnosis of very early SSc and requires further assessments [17].
The utility of nail-fold capillaroscopy as a tool to confirm the suspicion is also well established [30,31].

Implications
Rp is a common condition in the general population and the identification of patients with Rp who are at higher risk of developing a secondary form is still a challenge.This represents an early opportunity to identify patients likely to develop a CTD, with important consequences for management and monitoring strategies.This review adds further data to the existing literature; the work-up provided more comprehensive data than those reported in previous meta-analysis [9], thus we can hypothesize our estimations are more accurate.In conclusion, in consideration of the significant prognostic difference between what we considered pRp versus ssRp, these data relaunch the great debate about the classification of patients with Rp and only ANA positivity (immune signature) or only capillary abnormalities (vascular signature).Further prospective studies and a broader consensus appear essential.

Disclosure statement
No financial support and no other potential conflict of interest relevant to this article was reported.

Figure 1 :
Figure 1: Flow diagram of the systematic review process.

Figure 2A :
Figure 2A: Predictors of transition to connective tissue diseases in patients with Raynaud's phenomenon with immune signature.

Figure 2B :
Figure 2B: Predictors of transition to connective tissue diseases in patients with Raynaud's phenomenon with vascular signature.

Figure S1 :
Figure S1: Forest plots showing the incidence rate of transition to systemic sclerosis of primary and suspected secondary Raynaud's phenomenon as per 100 person-years.

Figure S2 :
Figure S2: Connective tissue diseases developed in patients with primary Raynaud's phenomenon (white column) and suspected secondary Raynaud's phenomenon (black column).

Table 1 :
Characteristics of the seven studies included and the quality of evidence assessment.

Table 2 :
Summary of findings on primary Raynaud's phenomenon (A) and on suspected secondary Raynaud's phenomenon (B).