Rheumatoid arthritis (RA) is one of the most prevalent systemic inflammatory diseases which involve joints and extra-articular tissues, thereby causing organ damage. Based on the chronic inflammation and immune dysregulation, the presence of RA has been associated with cardiovascular (CV) disease and an increased CV associated mortality [1, 2]. Classical risk factors for CV disease have been investigated in RA patients, however, epidemiological studies indicate that they cannot provide a sufficient explanation for the poorer CV prognosis of RA patients as compared to non-RA subjects [3, 4]. Therefore, a combination of yet not fully elucidated factors and regulatory mechanisms may contribute to CV morbidity in RA subjects. However, it is of clinical importance to identify markers which indicate an increased CV risk or predict CV mortality in RA patients. In this regard, recent studies reported an association of elevated red cell distribution width (RDW) with CV disease in the general population [5] but also in patients with RA [6, 7].

RDW is an automated measure of the range of variation of red blood cell (RBC) volume and is calculated as the standard deviation (SD) in red blood cell size divided by the mean corpuscular volume (MCV) (RDW (%) = 1 SD of RBC volume/MCV × 100). RDW is part of the complete blood count and has traditionally been used in anemia diagnosis [8–10] and to predict the response to iron treatment [11]. A retrospective analysis of > 20.000 patients with RA indicated that higher RDW, as well as increased levels of markers of inflammation, like C-reactive protein (CRP) or erythrocyte sedimentation rate (ESR), were associated with an increased risk of a subsequent CV event in RA-patients [12].

The mechanism behind the association of elevated RDW and CV risk in RA is yet incompletely understood. Possible explanations include that RDW reflects endothelial damage and impaired vascular repair, but also that RDW mirrors vascular inflammation underlying atherosclerosis and thereby effects myocardial infarction and stroke [13]. The described positive association between IFN-alpha, a cytokine contributing to endothelial damage, and RDW would be in agreement with this concept [14].

Of note, RDW is influenced by multiple factors related to erythropoiesis, such as iron, vitamin B12 or folic acid availability, as well as by hemolysis [9]. Moreover, RDW is also affected by organ dysfunctions (e.g. liver or renal dysfunction), inflammatory activity and some specific medications [15–17]. The latter might also affect the diagnostic potential of RDW in RA patients, as these subjects are treated with numerous disease modifying antirheumatic drugs (csDMARDs), like Methotrexate (MTX), or biological DMARDs (bDMARDs), including several cytokine antibodies [18]. According to EULAR recommendations, treatment of RA should be initiated with csDMARDs, most notably MTX in combination with low dose glucocorticoids. Although low dose MTX therapy is regarded as an anchor therapy in RA, full details of its mechanism of action and off target effects are still incompletely understood [19].

Because, MTX is well known to affect folic acid metabolism, MTX treatment can result in alterations of MCV, which may impact on RDW, as MCV levels feed into RDW calculation [6]. We thus questioned, whether RDW levels and subsequently its diagnostic utility and potential in RA subjects, as reported before [7, 11, 19, 20], are influenced by ongoing MTX therapy.

We then studied whether or not MTX treatment had an effect on the predictive value of RDW for severe CV events. The last routine follow up of patients with evaluation of laboratory parameters prior to the CV events was 82 days (mean) prior to such a CV complication. No significant differences in RDW, hemoglobin levels, CRP or ΔRDW (change in RDW between initial diagnosis and follow up prior to the CV event/last patient visit) were found in our cohort between patients incurring a CV event or not. Of note, we found a highly significant difference in RDW levels at last follow up, see definition above, in subjects without MTX treatment which was absent in subjects under MTX treatment comparing patients with and without a CV event (Fig. 1). Moreover, MTX-treated patients had significantly higher RDW levels than patients without MTX therapy (Table 4).

	MTX-use		p	CV-Event		p
	Yes (n = 222)	No (n = 163)		Yes (n = 23)	No (n = 362)
RDW, mean(SD), %	14,5 (1,4)	14,1 (1,6)	< 0,001	15,05 (2,2)	14,25 (1,4)	0,116
Hb, mean(SD), g/l	134,8 (14,5)	132,9 (14)	0,101	130,2 (14,4)	134,2 (14,3)	0,058
CRP, mean(SD), mg/dl	0,65 (0,97)	1,03 (2,1)	0,559	0,99 (1,1)	0,8 (1,6)	0,060
ΔRDW, mean(SD)	0,53 (1,24)	0,23 (1,0)	0,057	0,45 (0,6)	0,43 (1,2)	0,809

Effects of MTX intake on laboratory parameters at last visit prior to the CV event/last follow up. ΔRDW Change in RDW between initial diagnosis and follow up prior to the CV event/last follow up in patients without CV event. Hb hemoglobin level. In the overall cohort RDW at established disease was not applicable as predictive marker for a CV event. A tendency with lower hemoglobin levels and higher CRP was shown. MTX intake significantly affected RDW

In patients without MTX therapy RDW was significantly higher in those with a subsequent CV event. (p = 0,006; Fig. 1). This predictive value of RDW was abolished in patients taking MTX (p = 0,448, Fig. 1).

When performing binary regression for the risk of a CV event, we found that MTX naïve patients had a significant correlation between RDW and a CV event (Table 5B) which was not true for patients receiving MTX.

As far as we know, this is the first study evaluating the influence of csDMARDs on RDW. MTX impacts on RDW hence on the predictive value of RDW for CV events. Previous studies suggest that RDW is a good prognostic marker for CV disease and survival, but none of them evaluated concomitant treatment [6, 7, 12, 17]. This was also confirmed in our study indicating that an enhanced RDW at initial diagnosis, but neither ACPA nor RF positivity, is associated with an increased risk of a severe CV event [26].

However, we found that the predictive potential of RDW during follow up largely depends on the treatment of patients. Specifically, we identified that the diagnostic value of RDW as a risk indicator for subsequent CV disease is abolished in RA patients receiving MTX therapy. Although MTX is regarded as the anchor drug in RA, the mechanism of action is incompletely understood and it has been associated with negative effects on hematopoiesis, mainly via its impact on folic acid pathways, but also via direct toxic effects on hematopoietic progenitors. Accordingly, MTX but neither other csDMARDs nor bDMARD treatment resulted in alterations of red blood cell volume (MCV) and haemoglobin content of erythrocytes [27, 28]. To avoid such negative effects, patients under MTX therapy are supplemented with folic acid which was also the case in our subjects under MTX treatment [29]. However, RA patients under MTX treatment had increased RDW and MCV levels as compared to RA patients without MTX. It remains to be clarified, whether this can be referred to effects of MTX not linked to folic acid deficiency, or a reduced compliance of patients in regard to folic acid supplementation, which we could not study, because incomplete results of folic acid determination in blood were available in our study cohort. However, while MTX treatment resulted in a loss of the predictive value of RDW for subsequent CV events, higher RDW at initial diagnosis of all patients, at follow up prior to a CV event in patients without MTX treatment were significantly associated with an increased risk for a CV event.

Our study has limitations because of the retrospective design, the low number of patients with CV events and the fact that we had only insufficient data to evaluate the association of RDW, MTX therapy and CV events with other important variables including classical CV factors, the implication of a genetic component, or iron homeostasis [11, 30, 31]. In this regard, patients with RA who carried the methylene tetrahydrofolate reductase (MTHFR) 1298 allele C frequency were previously found to have an increased frequency of CV events after 5 and 10 years of follow-up. Moreover, patients carrying the MTHFR 1298 AC and CC genotypes had a significantly decreased flow-mediated endothelium-dependent vasodilatation, a marker of endothelial dysfunction that is an early indicator of atherogenesis, when compared with those carrying the MTHFR 1298 AA genotype [32]. More recent results also indicate that MTHFR expression is significantly reduced in patients with RA compared to controls. It was found to be especially true for RA patients with ischemic heart disease [33]. Taken these considerations together, these results indicate that MTHFR gene may influence the risk of subclinical atherosclerosis and CV disease in patients with RA.

Therefore, we need prospective evaluations of such risk markers and profiles to gain further insights into the functional and diagnostic role of RDW and alterations of hematological parameters as risk predicators for CV events in patients with RA. Accordingly, such confirmation may also translate into clinical practice, because patients at a higher risk, based on increased RDW may deserve an intensified clinical follow up with an improved control of classical CV risk factors, such as lipid status, hypertension, smoking status or hyperuricemia.

Our study approves RDW at initial diagnosis of RA as a risk predictor for serious CV events but also indicates that its predictive value is lost during follow up in patients receiving MTX therapy, whereas it remains valid in subjects receiving non-MTX containing treatments.