- Email: [email protected]
Clinical meaning and implications of serum hemoglobin levels in patients with rheumatoid arthritis
Author’s Accepted Manuscript Clinical meaning and implications of serum hemoglobin levels in patients with rheumatoid arthritis Ivan Padjen, Leopold Öhler, Paul Studenic, Thasia Woodworth, Josef Smolen, Daniel Aletaha www.elsevier.com/locate/semarthrit
PII: DOI: Reference:
S0049-0172(17)30091-4 http://dx.doi.org/10.1016/j.semarthrit.2017.03.001 YSARH51157
To appear in: Seminars in Arthritis and Rheumatism Cite this article as: Ivan Padjen, Leopold Öhler, Paul Studenic, Thasia Woodworth, Josef Smolen and Daniel Aletaha, Clinical meaning and implications of serum hemoglobin levels in patients with rheumatoid arthritis, Seminars in Arthritis and Rheumatism, http://dx.doi.org/10.1016/j.semarthrit.2017.03.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Clinical meaning and implications of serum hemoglobin levels in patients with rheumatoid arthritis AUTHORS: Ivan Padjena,1, Leopold Öhlerb, Paul Studenica, Thasia Woodworthc, Josef Smolena,d, Daniel Aletahaa
AFFILIATIONS a
Division of Rheumatology, Department of Medicine 3, Medical University of Vienna,
Währinger Gürtel 18-20, 1090 Vienna, Austria b
Department of Internal Medicine I, St. Josef Hospital, Auhofstraße 189, 1130 Vienna,
Austria c
Division of Rheumatology, David Geffen School of Medicine, University of California, 200
UCLA Medical Plaza, Los Angeles, California, USA (previously Roche) d
Center for Rheumatic Diseases, 2nd Department of Medicine, Hietzing Hospital,
Wolkersbergenstraße 1, 1130 Vienna, Austria
1
CORRESPONDING AUTHOR:
Ivan Padjen, MD, PhD Permanent Address: Division of Clinical Immunology and Rheumatology, Department of Internal Medicine, University Hospital Centre Zagreb, Kispaticeva 12, 10 000 Zagreb, Croatia Phone: +385 1 2388330
Fax: +385 1 2388335
E-mail: [email protected]
1
ABSTRACT Objective: Anemia is a common problem in rheumatoid arthritis (RA), associated with radiographic progression and disability. We explored the association of hemoglobin with a comprehensive set of variables in RA patients. Methods: We included RA outpatients in the routine setting. For each patient we performed measurements (clinical measures, blood tests including serology, markers of acute phase response and iron metabolism, including hepcidin, and circulating hematopoietic precursor levels) at baseline and 12 weeks thereafter, and analyzed their changes in patients with a treatment adaptation at baseline. We performed principal component analysis (PCA) to identify thematic groups hemoglobin was related to. Then we constructed multivariable linear models to assess the contribution of individual variables to the variability of hemoglobin. Results: Eighty-eight patients were included (age: 58±12; disease duration: 9.3±9.6 years). Cross-sectionally (at baseline and week 12) hemoglobin levels were tied to iron metabolism and hematopoiesis, but not to clinical activity, based on thematic groups extracted from the PCA. In contrast, longitudinal changes in hemoglobin levels were closely linked to changes in clinical activity. Conversely, hepcidin reflected iron metabolism cross-sectionally, but changes in acute phase response longitudinally. In multivariable analysis variability components of hemoglobin were explainable by ferritin, ESR, evaluator global assessment (EGA) and iron levels, while components of hemoglobin changes were explained by changes in EGA mostly. Hepcidin was not independently associated with hemoglobin.
2
Conclusion: Besides its dependence on body iron status, changes in hemoglobin levels are strongly tied to disease activity, possibly revealing more about disease activity than other laboratory markers. KEY WORDS: rheumatoid arthritis; anemia; inflammation; hepcidin
ABBREVIATIONS: ACPA, anti-citrullinated protein antibodies; ACR, American College of Rheumatology; BFU-E, burst-forming unit; CBC, complete blood count; CDAI, clinical disease activity index; CFU-GEMM, common myeloid progenitor; CFU-GM, granulocyte/macrophage colony forming unit; CRP, C reactive protein; DAS, disease activity score; DAS28-ESR, 28-joint disease activity score based on the erythrocyte sedimentation rate; EGA, evaluator global assessment; ESR, erythrocyte sedimentation rate; EULAR, European League Against Rheumatism; HAQ, health assessment questionnaire; IL-6, interleukin 6; MCV, mean corpuscular volume; MST, morning stiffness; PCA, principal component analysis; PGA, patient global assessment; RA, rheumatoid arthritis; RF, rheumatoid factor; SD, standard deviation; SDAI, simplified disease activity index; SJC28, 28-swollen joint count; TJC28, 28tender joint count; TNF-α, tumor necrosis factor alpha; VAS, visual analogue scale.
3
1. INTRODUCTION Anemia is a common finding in patients with rheumatoid arthritis (RA) and has a multifactorial pathogenesis: it may be caused by pathologic iron homeostasis, impaired erythropoiesis, a blunted response to erythropoietin, or a combination of all (1). The role of anemia, or of hemoglobin levels in general, is therefore especially challenging in a chronic inflammatory disease like RA. This is a possible reason why hemoglobin levels, despite often being perceived to reflect the degree of systemic inflammation, i.e. being a negative marker of the acute phase response, are not included in RA composite activity indices, while the erythrocyte sedimentation rate (ESR) and the C reactive protein (CRP), are (2). In fact, the rather low correlation of hemoglobin with other variables related to RA disease activity was already observed during the validation of the original disease activity score (DAS) in 1992 (3). At the same time, anemia is a common finding in RA and a factor independently associated with physical disability (4); in addition, low hemoglobin levels have been found to be indicators of active clinical or subclinical inflammatory disease (5, 6). Consequently, hemoglobin levels are often considered by clinicians in their clinical management of RA, but due to their complexity, further clinical decisions are based on intuition because hemoglobin levels are not represented in major management recommendations for RA, e.g. as potential therapeutic (co-)targets in RA. Previous studies reported an increase in hemoglobin levels after starting targeted RA treatment (7-9), with some of the studies failing to report longitudinal changes in composite disease activity indices (7) or parameters routinely used in the assessment of iron metabolism (9). Furthermore, while an inverse correlation of chronic inflammation and the number of bone marrow erythroid cells in RA has been reported (10), this association has never been investigated longitudinally in patients with RA. 4
In the present study we aimed to assess the associations of hemoglobin levels with a comprehensive set of parameters covering clinical and serological status, the acute phase response, iron metabolism and hematopoietic constitution, in order to shed further light on the meaning of hemoglobin values in RA. For that purpose we used a prospective observational approach.
2. PATIENTS AND METHODS 2.1. Patients and follow-up We included RA patients who were seen regularly during routine control visits at the outpatient clinic of the Division of Rheumatology, Medical University of Vienna, a specialized academic rheumatologic center. Patients had to fulfill the 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) RA classification criteria to be enrolled in this study (11). Each patient was evaluated at baseline and 12 weeks thereafter. Patients were seen during their scheduled routine outpatient visits, and consented to storage and analysis of biomaterials (serum, urine) for research purposes in the context of a larger biobanking activity at our Department. Written informed consent was obtained from all patients according to the Declaration of Helsinki after approval by the Ethics Committee of the Medical University of Vienna.
5
2.2. Study assessments We collected demographic variables (age, disease and symptom duration) at baseline, and a number of clinical and hematological/biochemical variables at baseline and 12 weeks. Laboratory procedures have not changed over the inclusion or the follow-up period. Besides the routine core set measures of RA disease activity, peripheral blood tests were performed to assess levels of rheumatoid factor (RF; determined by nephelometry) and anticitrullinated protein antibodies (ACPA; determined by second generation anti-CCP tests), the complete blood count (CBC), ESR, CRP and serum levels of fibrinogen. The clinical disease activity index (CDAI), simplified disease activity index (SDAI) and the 28-joint disease activity score based on the ESR (DAS28-ESR) were calculated. Besides determining standard serum levels of iron, transferrin, transferrin saturation, and ferritin, we measured levels of hepcidin, i.e. its 25-amino acid “mature” form, using isotope dilution micro-HPLC-tandem mass spectrometry, as previously described in detail (12). Hepcidin is induced in the liver by interleukin 6 (IL-6) as an acute phase response protein inhibiting intestinal iron absorption and release of iron from reticuloendothelial storages (13). The number of circulating hematopoietic progenitor cells was assessed as described previously (14). After a culture period of 14 days at 37°C in 5% CO2 and full humidity, cultures of peripheral blood mononuclear cells were examined under an inverted microscope. Aggregates with more than 40 translucent, compact, or dispersed cells were counted as granulocyte/macrophage colony forming units (CFU-GM). Bursts containing more than 100 hemoglobinized cells were counted as burst-forming units (BFU-E). Common myeloid progenitors (CFU-GEMM) were identified by their heterogeneous composition of 6
translucent and hemoglobinized cells. The number of colony-forming unit-cells (CFU-GM, BFU-E and CFU-GEMM) per milliliter of blood was calculated as described (14).
2.3. Statistical analyses We first analyzed all variables descriptively. Then we assessed changes in these variables between baseline and week 12 in the subgroup of patients with active disease and a treatment adaptation at baseline (CDAI≥10). Comparisons were performed using the Wilcoxon signed rank test. Principal component analysis (PCA) allows to summarize a large number of characteristics (variables) in a short list of factors (“latent” variables). The strength of the association of each original variable with the newly derived factors can be quantified by using loadings (ranging from 0 to 1, similar to correlation parameters; values over 0.400 are usually considered relevant (15)). As a next step, a “theme” can be attributed to each of the few new factors, according to variables most strongly loading to each factor. In other words, a factor with strong loadings from joint counts and global scores could be considered to represent the theme of “clinical activity”. The purpose of performing the PCA was to explore which themes hemoglobin levels most strongly relate to. By assessing the association of hemoglobin with these themes at baseline, we could address one key objective of our study, namely to understand the position of hemoglobin in the complex network of variables used to depict inflammation in RA. The PCA was then repeated in the subgroup of patients with active disease at baseline using change scores on their variables between baseline and week 12 as the basis for the PCA.
7
We then approached our study objective from a different analytical perspective using multivariable linear regression modelling by exploration of variance components. All variables that were significantly correlated with hemoglobin in a prior univariable analysis were included in the multivariable model. As before, one model was constructed for baseline values (all patients) and another one for change scores between week 0 and 12 (active patient subset). Statistical analyses were performed using IBM SPSS Statistics version 23.0 (Armonk, NY, USA).
3. RESULTS 3.1. Patient characteristics Eighty-eight patients (68 females, 77%) were included in the study (mean±SD age: 58.0±11.6 years; disease duration: 9.3±9.6 years); the subgroup of active patients with a new DMARD at baseline comprised 35 patients (30 females, 86%). Baseline hemoglobin was lower in the subgroup of active patients (mean±SD: 12.8±1.7 g/dL) compared to patients inactive at baseline (13.5±1.3 g/dL)(p=0.023, t=-2.320). The active subgroup receiving a new DMARD improved significantly in the CDAI, SDAI, DAS28-ESR, the 28-tender joint count (TJC28), patient global assessment (PGA), evaluator global assessment (EGA), patient pain assessed by visual analogue scale (VAS), ESR and RF level (p<0.05)(Table 1). A tendency towards improvement in levels of CRP, hepcidin, platelet counts, transferrin and ferritin, as well as duration of morning stiffness (MST) was also observed (p<0.10).
8
3.2. Themes extracted by principal component analysis The PCA identified four major factors using the data at baseline and week 12 steady state for all patients, as well as the data on the 12-week score changes for patients who were active at baseline. Based on the variables with the strongest loading to each of these factors, we attributed the following themes to the four factors: “clinical activity”, “acute phase response”, “iron metabolism” and “hematopoiesis”. In the steady state at baseline and week 12, hemoglobin levels were primarily ascribable to “iron metabolism” (Table 2; loading 0.647 at baseline and 0.541 at week 12) and to “hematopoiesis” (0.457 at baseline and 0.646 at week 12), while they were only very moderately loading on the factor “clinical activity” (0.139 at baseline and 0.117 at week 12). Still, the loading of hemoglobin on “clinical activity” at baseline was higher than the loading of acute phase measures (ESR, CRP and fibrinogen) on this factor. “Acute phase response” was considered to represent a distinct theme given the strong loadings of ESR, CRP and fibrinogen (Table 2) on one of the identified factors. In summary, in a steady state, hemoglobin levels mainly represent iron metabolism and only to a very small extent clinical disease activity (Figure 1). Interestingly, when looking at changes between baseline and week 12 in patients active at baseline, changes in hemoglobin levels were loading on the change in “iron metabolism” only to a negligible extent (0.055). In fact, the change in hemoglobin was now loading most strongly on the “clinical activity” theme (0.688). These findings indicate a distinct contrast between the meaning of hemoglobin levels at the steady state and the meaning of their longitudinal changes.
9
Hepcidin, a protein closely involved in the regulation of iron metabolism, like hemoglobin, was loading on the theme of “iron metabolism” at baseline (0.499) and at 12 weeks (0.591). Longitudinally, hepcidin clearly loaded on the “acute phase response” theme (0.579), while it had no meaningful loading on “iron metabolism” (0.059), as had been observed in the cross-sectional analysis and as it would be expected from its known biological activity (Table 2). Hematopoiesis had little in common with the other themes, and the major variables determining this theme (CFU-GEMM, BFU-E, and CFU-GM) showed no loading on any other theme in both the steady state and the longitudinal analysis.
3.3. Explanatory components of hemoglobin levels At baseline, hemoglobin was negatively correlated with the CDAI (-0.216), SDAI (-0.215), DAS28-ESR (-0.265), the 28-swollen joint count (SJC28) (-0.222), EGA (-0.244), ESR (-0.313), as well as the platelet count (-0.272), and positively correlated with the erythrocyte count (0.656), hematocrit (0.949), the mean corpuscular volume (MCV) (0.303), iron levels (0.401), transferrin saturation (0.349), ferritin levels (0.322), BFU-E (0.343) and CFU-GM (0.303) (p<0.05). Evaluation of the change in hemoglobin from baseline to week 12 revealed a negative correlation with the baseline hemoglobin (-0.378) indicating that patients with anemia of chronic disease who are effectively treated for RA are more likely to improve their hemoglobin levels, while those with normal hemoglobin levels at baseline show less improvement. This is also supported by a negative correlation of hemoglobin changes with changes in disease activity measures, such as the CDAI (-0.426), SDAI (-0.504), DAS28-ESR (0.448), SJC28 (-0.481), EGA (-0.452), ESR (-0.439), CRP (-0.581), or fibrinogen (-0.535). Less 10
surprisingly, we observed a positive correlation of hemoglobin changes and changes in the erythrocyte count (0.713), hematocrit (0.930), MCV (0.363), iron (0.579) and transferrin saturation (0.511) (p<0.05). However, when we introduced all variables that were significant in the correlation analysis into a multivariate analysis explaining variance components of hemoglobin, we found contributions of EGA, ESR, iron, and ferritin to the explanation of baseline hemoglobin levels (R2=0.407, p=0.033; Figure 2, left panel). Change in hemoglobin was mostly explained by change in EGA and less pronouncedly by change in serum iron levels (R2=0.493, p=0.034; Figure 2, right panel). Since the link between hemoglobin and EGA might be considered as circular, i.e. changes in hemoglobin levels might have been interpreted towards reflecting disease activity, we repeated this analysis and excluded EGA. The results revealed that fibrinogen, i.e. a marker reflecting the acute phase response, was the next single strongest contributor to changes in hemoglobin levels (total R2=0.293, p=0.002; data not shown). The detailed percentages of hemoglobin variability explained by each variable at baseline and 12-week changes are depicted on Figure 2. Of note, erythrocyte counts and hematocrit levels were not included in the multivariate analysis to avoid multicollinearity and direct biological circularity; composite disease activity indices were also not included given that they are mathematical constructs of single variables already included in the multivariate analysis.
4. DISCUSSION Although the individual level of hemoglobin only partly reflects the level of RA disease activity in the steady state, its changes are closely, though inversely, linked to changes in the 11
acute phase response, and even stronger to changes in clinical measures of disease activity. Indeed, in our multivariable model the change in EGA was the variable explaining the largest proportion of variability of hemoglobin. Although we have not assessed predictors of EGA, it is unlikely to be driven by anemia in our patients, since the biometricians responsible for performing the global assessment at our unit do not have laboratory test results at hand when making their judgement about each patient’s global status. This was also confirmed in an additional model, in which we excluded EGA and it became replaced by fibrinogen, a laboratory parameter loading both on disease activity and acute phase response in the PCA. A finding similar to ours was observed in the study by Hashimoto et al. who identified change in CDAI as a predictor of hemoglobin change (9). At a given time point, such as the baseline and week 12 in our study, however, hemoglobin levels are strongly tied to iron metabolism and hematopoiesis. In other words, the mere presence of anemia in a patient with RA does not necessarily indicate active disease, and vice versa. The link between clinical disease activity and anemia bears an important clinical implication, as it has been shown that anemic RA patients, independent of their disease activity, experience faster radiographic progression compared to their non-anemic counterparts (5, 6). It has been argued that there is a possible role of anemia as a more subtle surrogate for active RA. At the same time, anemia is also indirectly reflected in the ESR, which increases as the erythrocyte count declines (16). Laboratory variables routinely used to assess the level of systemic inflammation in patients with RA are the ESR and CRP and, less frequently, platelet counts and fibrinogen (17). All of these four variables are ascribable to the same theme in our analysis that we therefore labeled ”acute phase response“. Anemia of chronic disease usually develops in an
12
environment of systemic inflammation, i.e. the same context as the acute phase response. Despite being perceived in this context, hemoglobin is not defined as a variable of acute phase response sensu stricto. Both hemoglobin and variables of iron metabolism (iron – less reliable due to its fluctuations; ferritin; transferrin; and transferrin saturation) have been a part of the extended laboratory evaluation of patients with chronic inflammatory conditions such as RA (18). Indeed, ferritin and transferrin act as acute phase response proteins (19, 20), but that role is probably less prominent compared to their role as markers of iron metabolism: since ferritin and transferrin saturation were ascribable to a single theme at baseline - different from the “acute phase response” theme - we labeled that theme “iron metabolism”. Recognition of hepcidin as a link between systemic inflammation and reduction of iron availability has filled an important gap in the understanding of the pathogenesis of anemia of chronic disease including anemia of chronic inflammation in RA (21). However, the role of hepcidin exceeds the context of inflammation, since it is induced in hepatocytes in the state of iron overload, regulating the amount of total body iron (22). In the context of iron overload, the function of hepcidin is independent of IL-6, a key cytokine involved in inflammation and activation of the acute phase response. Therefore, from this mechanistic perspective, hepcidin can be perceived as acting at a crossroad between systemic inflammation and iron homeostasis (21, 23). Previous studies revealed a positive crosssectional association of hepcidin with markers of iron metabolism (primarily ferritin) and acute phase reactants such as CRP, ESR and haptoglobin (7, 8, 21, 24). Indeed, in our study, a dual role of hepcidin was observed using PCA, revealing high loadings on the “iron metabolism” theme at a steady state, as well as on the “acute phase response” theme for its change scores after initiation of therapy. Not only does the hepcidin level increase in the 13
setting of inflammation, which is an action driven by IL-6 as the necessary and sufficient cytokine (25), but it also rapidly decreases (within days) if IL-6 activity goes down. The latter effect was observed with the administration of tocilizumab, an IL-6 receptor blocker (7-8). IL-6 receptor blockade seems to be superior to tumor necrosis factor alpha (TNF-α) inhibition in lowering hepcidin and improving hemoglobin levels (8). This finding may support the role of IL-6 as a pivotal cytokine driving anemia of RA. Despite the associations observed in the PCA setting, baseline hepcidin and its 12-week changes were not correlated with hemoglobin in univariable analysis. In the multivariable models, baseline hemoglobin was associated with variables of iron metabolism other than hepcidin (Figure 2, left panel). A lacking direct association of hepcidin and hemoglobin has been reported both cross-sectionally and longitudinally (7, 24). Knowing that hepcidin plays a role in the pathogenesis of anemia of chronic disease/inflammation, failure to identify an independent association between hemoglobin and hepcidin could be explained by a more complex link between inflammation and anemia where the hepcidin-mediated mechanism may not be dominant, or at least not direct (7). There is a growing knowledge of the potential effect of pro-inflammatory cytokines, such as TNF-α, IL-1 and interferon-γ on bone marrow suppression (26, 27), in vitro and in mouse models. Furthermore, TNF-α has been associated with the apoptotic depletion of hematopoietic precursor cells in patients with RA (28), illustrating its direct effect on the development of anemia, bypassing the previously described mechanism involving hepcidin. Nonetheless, hematopoietic precursor cell ex vivo cultures did not fill the gap of unexplained hemoglobin variability in either of the two models assessed in our study:
14
neither RA disease activity nor the extent of anemia was reflected by changes in the progenitor cell levels. Our study has several limitations. PCA is used for exploratory purposes. Loadings by PCA do not prove any biological association. The methods used in our study served the purpose to increase our ability to interpret the meaning and place of hemoglobin in the complex network of variables used to describe inflammation and anemia of RA. For this reason we did not include healthy individuals in this study. A further limitation may be the short followup period, which certainly allowed us to investigate the associations of change scores, but which was at the same time too short to assess long-term effects on harder outcomes, such as radiographic progression. Also, although we are confident that we have covered all possible domains in relation to anemia (and its causes), we were not able to assess the impact of cytokines or to adjust for comorbidities.
5. CONCLUSIONS In summary, this analysis allowed us to obtain an unprecedented, multi-faceted, crosssectional and short-term longitudinal view of the meaning of hemoglobin in RA. In addition to its dependence on body iron status, it seems that hemoglobin may act as a disease activity measure, being at least as informative as “classical” variables associated with the acute phase response, if not more informative. In the light of all these associations, a large proportion of the variability of hemoglobin values still remains unexplainable. Despite the well-described regulatory role of hepcidin, its association with anemia in RA is not directly apparent from our analyses, and remains to be fully understood.
15
Further studies may be warranted to assess the value of using serum hemoglobin levels to interpret a patient’s inflammatory status and even to guide treatment decisions, especially when compared to the role of parameters used in the standard assessment of disease activity in RA.
CONFLICTS OF INTEREST: All authors declare that they have no conflict of interest. Thasia Woodworth was previously the Tocilizumab Clinical Science leader for Roche and contributed to study design and conduct. Roche performed the measurement of hepcidin. This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.
REFERENCES 1. Weiss G, Goodnough LT. Anemia of chronic disease. N Engl J Med 2005; 352: 1011–23. 2. Smolen JS, Aletaha D. Scores for all seasons: SDAI and CDAI. Clin Exp Rheumatol 2014; 32(5 Suppl 85): S–75–9. 3. van der Heijde DM, van’t Hof MA, van Riel PL, van Leeuwen MA, van Rijswijk MH, van de Putte LB. Validity of single variables and composite indices for measuring disease activity in rheumatoid arthritis. Ann Rheum Dis 1992; 51: 177–81. 4. Han C, Rahman MU, Doyle MK et al. Association of anemia and physical disability among patients with rheumatoid arthritis. J Rheumatol 2007; 34: 2177–82.
16
5. Möller B, Scherer A, Förger F, Villiger PM, Finckh A; Swiss Clinical Quality Management Program for Rheumatic Diseases. Anaemia may add information to standardised disease activity assessment to predict radiographic damage in rheumatoid arthritis: a prospective cohort study. Ann Rheum Dis 2014; 73: 691–6. 6. van Steenbergen HW, van Nies JA, van der Helm-van Mil AH. Anaemia to predict radiographic progression in rheumatoid arthritis. Ann Rheum Dis 2013; 72: e16. 7. Isaacs JD, Harari O, Kobold U, Lee JS, Bernasconi C. Effect of tocilizumab on haematological markers implicates interleukin-6 signalling in the anaemia of rheumatoid arthritis. Arthritis Res Ther 2013; 15: R204. 8. Song SN, Iwahashi M, Tomosugi N et al. Comparative evaluation of the effects of treatment with tocilizumab and TNF-α inhibitors on serum hepcidin, anemia response and disease activity in rheumatoid arthritis patients. Arthritis Res Ther 2013; 15: R141. 9. Hashimoto M, Fujii T, Hamaguchi M et al. Increase of hemoglobin levels by anti-IL-6 receptor antibody (tocilizumab) in rheumatoid arthritis. PloS One 2014; 9: e98202. 10. Papadaki HA, Kritikos HD, Valatas V, Boumpas DT, Eliopoulos GD. Anemia of chronic disease in rheumatoid arthritis is associated with increased apoptosis of bone marrow erythroid cells: improvement following anti-tumor necrosis factor-alpha antibody therapy. Blood 2002; 100: 474–82. 11. Aletaha D, Neogi T, Silman AJ et al. 2010 Rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Arthritis Rheum 2010; 62: 2569–81.
17
12. Kobold U, Dülffer T, Dangl M et al. Quantification of hepcidin-25 in human serum by isotope dilution micro-HPLC-tandem mass spectrometry. Clin Chem 2008; 54: 1584-6. 13. Raj DS. Role of interleukin-6 in the anemia of chronic disease. Semin Arthritis Rheum 2009; 38: 382–8. 14. Sagaster V, Jäger E, Weltermann A et al. Circulating hematopoietic progenitor cells predict survival in patients with myelofibrosis with myeloid metaplasia. Haematologica 2003; 88: 1204–12. 15. Field A. Exploratory factor analysis. In Field A: Discovering statistics using SPSS. 2nd ed., Los Angeles-London-New Delhi-Singapore-Washington DC, Sage 2009: 627–85. 16. Kanfer EJ, Nicol BA. Haemoglobin concentration and erythrocyte sedimentation rate in primary care patients. J R Soc Med 1997; 90: 16–8. 17. Scott DL, Wolfe F, Huizinga TW. Rheumatoid arthritis. Lancet 2010; 376: 1094–108. 18. Osterholm EA, Georgieff MK. Chronic Inflammation and Iron Metabolism. J Pediatr 2015; 166: 1351–7.e1. 19. Gruys E, Toussaint MJM, Niewold TA, Koopmans SJ. Acute phase reaction and acute phase proteins. J Zhejiang Univ Sci 2005; 6B: 1045–56. 20. Yildirim K, Karatay S, Melikoglu MA, Gureser G, Ugur M, Senel K. Associations between acute phase reactant levels and disease activity score (DAS28) in patients with rheumatoid arthritis. Ann Clin Lab Sci 2004; 34: 423–6.
18
21. Demirag MD, Haznedaroglu S, Sancak B et al. Circulating hepcidin in the crossroads of anemia and inflammation associated with rheumatoid arthritis. Intern Med 2009; 48: 421–6. 22. Singh B, Arora S, Agrawal P, Gupta SK. Hepcidin: a novel peptide hormone regulating iron metabolism. Clin Chim Acta 2011; 412: 823–30. 23. Choy E. Understanding the dynamics: pathways involved in the pathogenesis of rheumatoid arthritis. Rheumatol 2012; 51 Suppl 5: v3–11. 24. Sellam J, Kotti S, Fellahi S et al. Serum hepcidin level is not an independent surrogate biomarker of disease activity or of radiographic progression in rheumatoid arthritis: results from the ESPOIR cohort. Ann Rheum Dis 2013; 72: 312–4. 25. Nemeth E, Rivera S, Gabavan V et al. IL-6 mediates hypoferremia of inflammation by inducing the synthesis of the iron regulatory hormone hepcidin. J Clin Invest 2004; 113: 1271-6. 26. Libregts SF, Gutiérrez L, de Bruin AM et al. Chronic IFN-γ production in mice induces anemia by reducing erythrocyte life span and inhibiting erythropoiesis through an IRF-1/PU.1 axis. Blood 2011; 118: 2578–88. 27. Nemeth E, Ganz T. Anemia of inflammation. Hematol Oncol Clin North Am 2014; 28: 671– 81, vi. 28. Papadaki HA, Kritikos HD, Gemetzi C et al. Bone marrow progenitor cell reserve and function and stromal cell function are defective in rheumatoid arthritis: evidence for a tumor necrosis factor alpha-mediated effect. Blood 2002; 99: 1610–9.
19
TABLES Table 1. Clinical measures and laboratory variables measured at baseline and week 12. Parameter Baseline alla Baseline activeb Week 12 activec CDAI 6.9 (0.85-15.48) 17.5 (12.8-24.8) 12.3 (6.8-22.6) SDAI 7 (1.35-16.57) 20.7 (13.88-26.48) 13.58 (7.37-24) DAS28-ESR 2.97 (2.1-4.01) 4.53 (3.68-5.39) 3.49 (2.58-4.92) SJC28 1 (0-4) 4 (3-7) 4 (2-6) TJC28 0 (0-2) 4 (2-10) 2 (0-5) PGA 23 (3.25-49.75) 50 (39-64) 40 (14-50) EGA 12 (0-25.75) 28 (22-37) 20.5 (13-33.5) Pain (VAS) 20 (3-43.75) 44 (31-63) 32 (17-45) HAQ 0.25 (0-1) 1 (0.38-1.38) 0.63 (0.38-1) MST (minutes) 0 (0-27.5) 30 (0-60) 10 (0-52.6) RF (U/L) 25 (12-129) 51.05 (12-209.25) 33 (12-169) ACPA (U/L) 103 (2.1-600) 154 (1.6-600) 132 (1.5-600) ESR (mm/h) 16 (9-25) 17 (9.75-26) 15 (8-22.5) CRP (mg/dL) 0.44 (0.17-0.98) 0.58 (0.25-2.24) 0.39 (0.19-0.91) Fibrinogen (mg/dL) 397 (340.5-482.25) 408 (357.75-487.25) 381 (349-459) Hepcidin (nmol/L) 4.65 (1.03-12.43) 6.72 (1.05-14.75) 3.21 (0.51-10.81) Erythrocytes (x10E12/L) 4.4 (4.2-4.7) 4.4 (4.1-4.7) 4.5 (4.2-4.8) Hemoglobin (g/dL) 13.2 (12.4-14.1) 12.8 (11.8-14.1) 13.1 (12.2-13.9) Hematocrite (%) 39.8 (37.6-42.1) 39.3 (36-41.8) 39.5 (36.7-41.6) MCV (fL) 90.2 (85.8-93.3) 88.9 (82.3-93.2) 89 (84.4-92.2) Reticulocytes (x10E9/L) 41.3 (33.7-59.05) 42.7 (30.9-65.2) 45.55 (35.6-62.73) Leukocytes (x10E9/L) 7.78 (6.11-9.85) 8.5 (6.36-11.1) 7.64 (5.59-11.08) Platelets (x10E9/L) 260 (222-295) 270 (236-312) 259 (226-289) Iron (ug/dL) 71 (51-92.75) 64 (37-89) 61 (41-99) Transferrin (ug/dL) 257.65 (232.83-287.3) 256.2 (227.6-287.3) 260.5 (240.6-293.2) Transferrin saturation (%) 19.8 (13.38-27.4) 18.3 (11.3-25.7) 18 (10.3-31.4) Ferritin (ng/mL) 55.95 (29.8-94.88) 52.4 (30.2-91.2) 38.8 (25.5-92.8) BFU-E (ml-1) 641 (410-1004) 636 (481-916) 626.5 (325.5-968) -1 CFU-GM (ml ) 86 (53.25-139.75) 91 (65-132) 70 (36.5-145.25) -1 CFU-GEMM (ml ) 11 (0-22) 11 (3-21) 12 (4-15.25) a values at beaseline for all patients (left); bvalues at baseline for initially active patients (centre); cvalues at week 12 for initially active patients (right); *baseline active and week 12 active compared (Wilcoxon signed ranks test (2-tailed)); data are presented as medians and interquartile ranges.
20
p* 0.009 0.006 0.004 0.415 0.003 0.003 0.013 0.002 0.212 0.058 0.007 0.147 0.024 0.057 0.158 0.083 0.162 0.174 0.993 0.222 0.517 0.497 0.064 0.478 0.062 0.774 0.092 0.778 0.200 0.333
Table 2. Principal component analysis – loadings of variables to each theme. Loadings presented at baseline and week 12 (for all patients), as well as at changes between baseline and week 12 (for initially active patients)
Included
Signatu re* for
CDAI
(1)
SDAI DAS28ESR
(1) (1)
SJC28 TJC28 EGA PGA HAQ Pain (VAS) MST RF CCP ESR
(3)
CRP Fibrinoge n Erythrocy tes Hemoglo bin Hematocr it
(3)
MCV Reticuloc ytes Leukocyt
(3)
(1) Clinical (2) (4) Iron activity Hematopoiesis (3) Acute phase metabolism BL 12 Δ BL 12 Δ BL 12 Δ BL 12 0.9 0.9 1.0 0.0 0.0 0.0 0.0 0.0 0.1 0.1 0.0 68 98 00 71 29 01 04 91 31 05 00 0.9 0.9 0.9 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 36 64 10 85 38 28 10 58 96 71 09 0.8 0.8 0.8 0.1 0.1 0.0 0.2 0.1 0.0 0.0 0.0 01 42 54 90 71 25 07 89 46 26 69 0.7 0.6 0.4 0.2 0.0 0.0 0.1 0.0 0.4 0.0 0.2 42 74 51 04 80 38 44 53 06 97 65 0.7 0.7 0.9 0.0 0.0 0.0 0.0 0.2 0.3 0.1 0.2 46 86 39 07 32 55 03 00 27 99 38 0.8 0.8 0.9 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.1 85 48 00 69 32 96 84 52 13 94 72 0.7 0.8 0.3 0.0 0.0 0.1 0.2 0.0 0.1 0.1 0.0 88 42 25 15 53 66 23 74 70 03 69 0.4 0.5 0.1 0.0 0.1 0.1 0.1 0.0 0.1 0.0 0.3 18 61 54 89 37 60 25 74 06 65 01 0.7 0.7 0.1 0.0 0.0 0.2 0.2 0.0 0.3 0.0 0.0 85 85 29 52 99 06 09 12 88 97 29 0.6 0.5 0.2 0.0 0.1 0.1 0.1 0.1 0.2 0.1 0.0 23 27 23 43 87 84 98 19 34 80 66 0.1 0.0 0.0 0.2 0.1 0.0 0.0 0.1 0.0 0.1 0.0 17 84 63 27 58 57 18 37 62 35 69 0.1 0.2 0.1 0.3 0.2 0.0 0.0 0.1 0.0 0.0 0.0 62 81 91 11 05 86 64 74 68 53 82 0.1 0.0 0.1 0.4 0.1 0.1 0.5 0.8 0.5 0.2 0.0 01 56 03 36 78 54 58 02 12 20 14 0.1 0.0 0.0 0.1 0.1 0.0 0.7 0.6 0.9 0.1 0.0 20 38 06 22 20 28 19 92 08 93 47 0.0 0.1 0.4 0.1 0.1 0.0 0.6 0.6 0.5 0.4 0.1 83 70 58 19 59 70 46 40 64 29 95 0.0 0.0 0.2 0.6 0.7 0.1 0.0 0.0 0.4 0.4 0.0 36 22 08 55 43 78 89 20 55 02 43 0.1 0.1 0.6 0.4 0.6 0.1 0.3 0.2 0.1 0.6 0.5 39 17 88 57 46 51 96 53 69 47 41 0.0 0.0 0.6 0.5 0.6 0.1 0.3 0.2 0.3 0.6 0.4 96 82 07 17 58 57 59 29 11 17 54 0.2 0.0 0.0 0.1 0.1 0.2 0.4 0.2 0.6 0.0 0.6 34 82 28 18 74 47 46 81 71 21 57 0.2 0.0 0.1 0.0 0.4 0.6 0.0 0.0 0.1 0.1 0.1 70 75 50 95 35 72 31 47 11 21 20 0.1 0.0 0.0 0.3 0.2 0.4 0.4 0.6 0.0 0.0 0.0
Δ 0.0 25 0.1 03 0.2 09 0.3 08 0.0 07 0.1 33 0.3 10 0.5 33 0.3 92 0.5 01 0.7 34 0.3 45 0.0 78 0.0 32 0.1 53 0.0 56 0.0 55 0.0 33 0.0 45 0.1 36 0.2 21
es Platelets Iron Transferri n Transf. Sat.
(4)
Ferritin
(4)
(4) (4)
Hepcidin BFU-E
(2)
CFU-GM CFUGEMM
(2) (2)
31 0.0 11 0.0 77 0.0 64 0.0 65 0.0 23 0.1 84 0.0 30 0.0 37 0.0 17
44 0.0 43 0.0 84 0.0 12 0.0 76 0.0 03 0.1 54 0.0 51 0.0 52 0.0 16
28 0.0 99 0.0 58 0.1 52 0.0 90 0.2 56 0.0 56 0.0 19 0.0 07 0.0 91
90 0.0 27 0.1 94 0.2 53 0.2 63 0.0 67 0.1 12 0.8 24 0.7 84 0.7 53
67 0.0 06 0.0 95 0.0 75 0.0 93 0.2 69 0.1 04 0.8 25 0.7 64 0.6 66
25 0.4 66 0.0 16 0.4 62 0.1 10 0.3 33 0.0 72 0.8 11 0.8 44 0.8 60
67 0.5 04 0.6 74 0.0 43 0.6 60 0.0 69 0.0 14 0.0 75 0.1 07 0.0 90
24 0.7 28 0.5 43 0.2 37 0.5 24 0.1 41 0.4 54 0.1 41 0.0 92 0.0 97
01 0.2 68 0.4 43 0.5 39 0.4 00 0.5 04 0.5 79 0.0 73 0.1 65 0.1 17
89 0.1 24 0.2 43 0.4 63 0.3 06 0.7 26 0.4 99 0.0 12 0.1 38 0.0 41
39 0.0 93 0.6 13 0.4 92 0.6 26 0.6 19 0.5 91 0.0 55 0.1 53 0.1 42
65 0.0 37 0.6 66 0.0 80 0.7 13 0.4 58 0.0 59 0.1 64 0.0 72 0.1 39
bold – loadings over 0.400; BL – baseline; 12 – week 12; Δ – changes between baseline and week 12; Transf. Sat. – transferrin saturation; *the variables with the highest loading leading to the labelling (“theme”) of a factor are indicated as signature variables for that respective factor.
22
FIGURES
Figure 1. Loadings of hemoglobin on the four themes. Full line: loadings at baseline (all patients); dotted line: loadings for changes between baseline and week 12 (patients active at baseline)
(note to the Editor: single-column fitting image)
23
Figure 2. Percentage of variability of hemoglobin explained by individual variables in two multivariable linear models. Baseline: model explaining the variability of baseline hemoglobin (all patients). Change scores: model explaining the variability of hemoglobin changes between baseline and week 12 (patients active at baseline). EGA – evaluator global assessment; ESR – erythrocyte sedimentation rate.
(note to the Editor: single-column fitting image)
24
PII: DOI: Reference:
S0049-0172(17)30091-4 http://dx.doi.org/10.1016/j.semarthrit.2017.03.001 YSARH51157
To appear in: Seminars in Arthritis and Rheumatism Cite this article as: Ivan Padjen, Leopold Öhler, Paul Studenic, Thasia Woodworth, Josef Smolen and Daniel Aletaha, Clinical meaning and implications of serum hemoglobin levels in patients with rheumatoid arthritis, Seminars in Arthritis and Rheumatism, http://dx.doi.org/10.1016/j.semarthrit.2017.03.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Clinical meaning and implications of serum hemoglobin levels in patients with rheumatoid arthritis AUTHORS: Ivan Padjena,1, Leopold Öhlerb, Paul Studenica, Thasia Woodworthc, Josef Smolena,d, Daniel Aletahaa
AFFILIATIONS a
Division of Rheumatology, Department of Medicine 3, Medical University of Vienna,
Währinger Gürtel 18-20, 1090 Vienna, Austria b
Department of Internal Medicine I, St. Josef Hospital, Auhofstraße 189, 1130 Vienna,
Austria c
Division of Rheumatology, David Geffen School of Medicine, University of California, 200
UCLA Medical Plaza, Los Angeles, California, USA (previously Roche) d
Center for Rheumatic Diseases, 2nd Department of Medicine, Hietzing Hospital,
Wolkersbergenstraße 1, 1130 Vienna, Austria
1
CORRESPONDING AUTHOR:
Ivan Padjen, MD, PhD Permanent Address: Division of Clinical Immunology and Rheumatology, Department of Internal Medicine, University Hospital Centre Zagreb, Kispaticeva 12, 10 000 Zagreb, Croatia Phone: +385 1 2388330
Fax: +385 1 2388335
E-mail: [email protected]
1
ABSTRACT Objective: Anemia is a common problem in rheumatoid arthritis (RA), associated with radiographic progression and disability. We explored the association of hemoglobin with a comprehensive set of variables in RA patients. Methods: We included RA outpatients in the routine setting. For each patient we performed measurements (clinical measures, blood tests including serology, markers of acute phase response and iron metabolism, including hepcidin, and circulating hematopoietic precursor levels) at baseline and 12 weeks thereafter, and analyzed their changes in patients with a treatment adaptation at baseline. We performed principal component analysis (PCA) to identify thematic groups hemoglobin was related to. Then we constructed multivariable linear models to assess the contribution of individual variables to the variability of hemoglobin. Results: Eighty-eight patients were included (age: 58±12; disease duration: 9.3±9.6 years). Cross-sectionally (at baseline and week 12) hemoglobin levels were tied to iron metabolism and hematopoiesis, but not to clinical activity, based on thematic groups extracted from the PCA. In contrast, longitudinal changes in hemoglobin levels were closely linked to changes in clinical activity. Conversely, hepcidin reflected iron metabolism cross-sectionally, but changes in acute phase response longitudinally. In multivariable analysis variability components of hemoglobin were explainable by ferritin, ESR, evaluator global assessment (EGA) and iron levels, while components of hemoglobin changes were explained by changes in EGA mostly. Hepcidin was not independently associated with hemoglobin.
2
Conclusion: Besides its dependence on body iron status, changes in hemoglobin levels are strongly tied to disease activity, possibly revealing more about disease activity than other laboratory markers. KEY WORDS: rheumatoid arthritis; anemia; inflammation; hepcidin
ABBREVIATIONS: ACPA, anti-citrullinated protein antibodies; ACR, American College of Rheumatology; BFU-E, burst-forming unit; CBC, complete blood count; CDAI, clinical disease activity index; CFU-GEMM, common myeloid progenitor; CFU-GM, granulocyte/macrophage colony forming unit; CRP, C reactive protein; DAS, disease activity score; DAS28-ESR, 28-joint disease activity score based on the erythrocyte sedimentation rate; EGA, evaluator global assessment; ESR, erythrocyte sedimentation rate; EULAR, European League Against Rheumatism; HAQ, health assessment questionnaire; IL-6, interleukin 6; MCV, mean corpuscular volume; MST, morning stiffness; PCA, principal component analysis; PGA, patient global assessment; RA, rheumatoid arthritis; RF, rheumatoid factor; SD, standard deviation; SDAI, simplified disease activity index; SJC28, 28-swollen joint count; TJC28, 28tender joint count; TNF-α, tumor necrosis factor alpha; VAS, visual analogue scale.
3
1. INTRODUCTION Anemia is a common finding in patients with rheumatoid arthritis (RA) and has a multifactorial pathogenesis: it may be caused by pathologic iron homeostasis, impaired erythropoiesis, a blunted response to erythropoietin, or a combination of all (1). The role of anemia, or of hemoglobin levels in general, is therefore especially challenging in a chronic inflammatory disease like RA. This is a possible reason why hemoglobin levels, despite often being perceived to reflect the degree of systemic inflammation, i.e. being a negative marker of the acute phase response, are not included in RA composite activity indices, while the erythrocyte sedimentation rate (ESR) and the C reactive protein (CRP), are (2). In fact, the rather low correlation of hemoglobin with other variables related to RA disease activity was already observed during the validation of the original disease activity score (DAS) in 1992 (3). At the same time, anemia is a common finding in RA and a factor independently associated with physical disability (4); in addition, low hemoglobin levels have been found to be indicators of active clinical or subclinical inflammatory disease (5, 6). Consequently, hemoglobin levels are often considered by clinicians in their clinical management of RA, but due to their complexity, further clinical decisions are based on intuition because hemoglobin levels are not represented in major management recommendations for RA, e.g. as potential therapeutic (co-)targets in RA. Previous studies reported an increase in hemoglobin levels after starting targeted RA treatment (7-9), with some of the studies failing to report longitudinal changes in composite disease activity indices (7) or parameters routinely used in the assessment of iron metabolism (9). Furthermore, while an inverse correlation of chronic inflammation and the number of bone marrow erythroid cells in RA has been reported (10), this association has never been investigated longitudinally in patients with RA. 4
In the present study we aimed to assess the associations of hemoglobin levels with a comprehensive set of parameters covering clinical and serological status, the acute phase response, iron metabolism and hematopoietic constitution, in order to shed further light on the meaning of hemoglobin values in RA. For that purpose we used a prospective observational approach.
2. PATIENTS AND METHODS 2.1. Patients and follow-up We included RA patients who were seen regularly during routine control visits at the outpatient clinic of the Division of Rheumatology, Medical University of Vienna, a specialized academic rheumatologic center. Patients had to fulfill the 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) RA classification criteria to be enrolled in this study (11). Each patient was evaluated at baseline and 12 weeks thereafter. Patients were seen during their scheduled routine outpatient visits, and consented to storage and analysis of biomaterials (serum, urine) for research purposes in the context of a larger biobanking activity at our Department. Written informed consent was obtained from all patients according to the Declaration of Helsinki after approval by the Ethics Committee of the Medical University of Vienna.
5
2.2. Study assessments We collected demographic variables (age, disease and symptom duration) at baseline, and a number of clinical and hematological/biochemical variables at baseline and 12 weeks. Laboratory procedures have not changed over the inclusion or the follow-up period. Besides the routine core set measures of RA disease activity, peripheral blood tests were performed to assess levels of rheumatoid factor (RF; determined by nephelometry) and anticitrullinated protein antibodies (ACPA; determined by second generation anti-CCP tests), the complete blood count (CBC), ESR, CRP and serum levels of fibrinogen. The clinical disease activity index (CDAI), simplified disease activity index (SDAI) and the 28-joint disease activity score based on the ESR (DAS28-ESR) were calculated. Besides determining standard serum levels of iron, transferrin, transferrin saturation, and ferritin, we measured levels of hepcidin, i.e. its 25-amino acid “mature” form, using isotope dilution micro-HPLC-tandem mass spectrometry, as previously described in detail (12). Hepcidin is induced in the liver by interleukin 6 (IL-6) as an acute phase response protein inhibiting intestinal iron absorption and release of iron from reticuloendothelial storages (13). The number of circulating hematopoietic progenitor cells was assessed as described previously (14). After a culture period of 14 days at 37°C in 5% CO2 and full humidity, cultures of peripheral blood mononuclear cells were examined under an inverted microscope. Aggregates with more than 40 translucent, compact, or dispersed cells were counted as granulocyte/macrophage colony forming units (CFU-GM). Bursts containing more than 100 hemoglobinized cells were counted as burst-forming units (BFU-E). Common myeloid progenitors (CFU-GEMM) were identified by their heterogeneous composition of 6
translucent and hemoglobinized cells. The number of colony-forming unit-cells (CFU-GM, BFU-E and CFU-GEMM) per milliliter of blood was calculated as described (14).
2.3. Statistical analyses We first analyzed all variables descriptively. Then we assessed changes in these variables between baseline and week 12 in the subgroup of patients with active disease and a treatment adaptation at baseline (CDAI≥10). Comparisons were performed using the Wilcoxon signed rank test. Principal component analysis (PCA) allows to summarize a large number of characteristics (variables) in a short list of factors (“latent” variables). The strength of the association of each original variable with the newly derived factors can be quantified by using loadings (ranging from 0 to 1, similar to correlation parameters; values over 0.400 are usually considered relevant (15)). As a next step, a “theme” can be attributed to each of the few new factors, according to variables most strongly loading to each factor. In other words, a factor with strong loadings from joint counts and global scores could be considered to represent the theme of “clinical activity”. The purpose of performing the PCA was to explore which themes hemoglobin levels most strongly relate to. By assessing the association of hemoglobin with these themes at baseline, we could address one key objective of our study, namely to understand the position of hemoglobin in the complex network of variables used to depict inflammation in RA. The PCA was then repeated in the subgroup of patients with active disease at baseline using change scores on their variables between baseline and week 12 as the basis for the PCA.
7
We then approached our study objective from a different analytical perspective using multivariable linear regression modelling by exploration of variance components. All variables that were significantly correlated with hemoglobin in a prior univariable analysis were included in the multivariable model. As before, one model was constructed for baseline values (all patients) and another one for change scores between week 0 and 12 (active patient subset). Statistical analyses were performed using IBM SPSS Statistics version 23.0 (Armonk, NY, USA).
3. RESULTS 3.1. Patient characteristics Eighty-eight patients (68 females, 77%) were included in the study (mean±SD age: 58.0±11.6 years; disease duration: 9.3±9.6 years); the subgroup of active patients with a new DMARD at baseline comprised 35 patients (30 females, 86%). Baseline hemoglobin was lower in the subgroup of active patients (mean±SD: 12.8±1.7 g/dL) compared to patients inactive at baseline (13.5±1.3 g/dL)(p=0.023, t=-2.320). The active subgroup receiving a new DMARD improved significantly in the CDAI, SDAI, DAS28-ESR, the 28-tender joint count (TJC28), patient global assessment (PGA), evaluator global assessment (EGA), patient pain assessed by visual analogue scale (VAS), ESR and RF level (p<0.05)(Table 1). A tendency towards improvement in levels of CRP, hepcidin, platelet counts, transferrin and ferritin, as well as duration of morning stiffness (MST) was also observed (p<0.10).
8
3.2. Themes extracted by principal component analysis The PCA identified four major factors using the data at baseline and week 12 steady state for all patients, as well as the data on the 12-week score changes for patients who were active at baseline. Based on the variables with the strongest loading to each of these factors, we attributed the following themes to the four factors: “clinical activity”, “acute phase response”, “iron metabolism” and “hematopoiesis”. In the steady state at baseline and week 12, hemoglobin levels were primarily ascribable to “iron metabolism” (Table 2; loading 0.647 at baseline and 0.541 at week 12) and to “hematopoiesis” (0.457 at baseline and 0.646 at week 12), while they were only very moderately loading on the factor “clinical activity” (0.139 at baseline and 0.117 at week 12). Still, the loading of hemoglobin on “clinical activity” at baseline was higher than the loading of acute phase measures (ESR, CRP and fibrinogen) on this factor. “Acute phase response” was considered to represent a distinct theme given the strong loadings of ESR, CRP and fibrinogen (Table 2) on one of the identified factors. In summary, in a steady state, hemoglobin levels mainly represent iron metabolism and only to a very small extent clinical disease activity (Figure 1). Interestingly, when looking at changes between baseline and week 12 in patients active at baseline, changes in hemoglobin levels were loading on the change in “iron metabolism” only to a negligible extent (0.055). In fact, the change in hemoglobin was now loading most strongly on the “clinical activity” theme (0.688). These findings indicate a distinct contrast between the meaning of hemoglobin levels at the steady state and the meaning of their longitudinal changes.
9
Hepcidin, a protein closely involved in the regulation of iron metabolism, like hemoglobin, was loading on the theme of “iron metabolism” at baseline (0.499) and at 12 weeks (0.591). Longitudinally, hepcidin clearly loaded on the “acute phase response” theme (0.579), while it had no meaningful loading on “iron metabolism” (0.059), as had been observed in the cross-sectional analysis and as it would be expected from its known biological activity (Table 2). Hematopoiesis had little in common with the other themes, and the major variables determining this theme (CFU-GEMM, BFU-E, and CFU-GM) showed no loading on any other theme in both the steady state and the longitudinal analysis.
3.3. Explanatory components of hemoglobin levels At baseline, hemoglobin was negatively correlated with the CDAI (-0.216), SDAI (-0.215), DAS28-ESR (-0.265), the 28-swollen joint count (SJC28) (-0.222), EGA (-0.244), ESR (-0.313), as well as the platelet count (-0.272), and positively correlated with the erythrocyte count (0.656), hematocrit (0.949), the mean corpuscular volume (MCV) (0.303), iron levels (0.401), transferrin saturation (0.349), ferritin levels (0.322), BFU-E (0.343) and CFU-GM (0.303) (p<0.05). Evaluation of the change in hemoglobin from baseline to week 12 revealed a negative correlation with the baseline hemoglobin (-0.378) indicating that patients with anemia of chronic disease who are effectively treated for RA are more likely to improve their hemoglobin levels, while those with normal hemoglobin levels at baseline show less improvement. This is also supported by a negative correlation of hemoglobin changes with changes in disease activity measures, such as the CDAI (-0.426), SDAI (-0.504), DAS28-ESR (0.448), SJC28 (-0.481), EGA (-0.452), ESR (-0.439), CRP (-0.581), or fibrinogen (-0.535). Less 10
surprisingly, we observed a positive correlation of hemoglobin changes and changes in the erythrocyte count (0.713), hematocrit (0.930), MCV (0.363), iron (0.579) and transferrin saturation (0.511) (p<0.05). However, when we introduced all variables that were significant in the correlation analysis into a multivariate analysis explaining variance components of hemoglobin, we found contributions of EGA, ESR, iron, and ferritin to the explanation of baseline hemoglobin levels (R2=0.407, p=0.033; Figure 2, left panel). Change in hemoglobin was mostly explained by change in EGA and less pronouncedly by change in serum iron levels (R2=0.493, p=0.034; Figure 2, right panel). Since the link between hemoglobin and EGA might be considered as circular, i.e. changes in hemoglobin levels might have been interpreted towards reflecting disease activity, we repeated this analysis and excluded EGA. The results revealed that fibrinogen, i.e. a marker reflecting the acute phase response, was the next single strongest contributor to changes in hemoglobin levels (total R2=0.293, p=0.002; data not shown). The detailed percentages of hemoglobin variability explained by each variable at baseline and 12-week changes are depicted on Figure 2. Of note, erythrocyte counts and hematocrit levels were not included in the multivariate analysis to avoid multicollinearity and direct biological circularity; composite disease activity indices were also not included given that they are mathematical constructs of single variables already included in the multivariate analysis.
4. DISCUSSION Although the individual level of hemoglobin only partly reflects the level of RA disease activity in the steady state, its changes are closely, though inversely, linked to changes in the 11
acute phase response, and even stronger to changes in clinical measures of disease activity. Indeed, in our multivariable model the change in EGA was the variable explaining the largest proportion of variability of hemoglobin. Although we have not assessed predictors of EGA, it is unlikely to be driven by anemia in our patients, since the biometricians responsible for performing the global assessment at our unit do not have laboratory test results at hand when making their judgement about each patient’s global status. This was also confirmed in an additional model, in which we excluded EGA and it became replaced by fibrinogen, a laboratory parameter loading both on disease activity and acute phase response in the PCA. A finding similar to ours was observed in the study by Hashimoto et al. who identified change in CDAI as a predictor of hemoglobin change (9). At a given time point, such as the baseline and week 12 in our study, however, hemoglobin levels are strongly tied to iron metabolism and hematopoiesis. In other words, the mere presence of anemia in a patient with RA does not necessarily indicate active disease, and vice versa. The link between clinical disease activity and anemia bears an important clinical implication, as it has been shown that anemic RA patients, independent of their disease activity, experience faster radiographic progression compared to their non-anemic counterparts (5, 6). It has been argued that there is a possible role of anemia as a more subtle surrogate for active RA. At the same time, anemia is also indirectly reflected in the ESR, which increases as the erythrocyte count declines (16). Laboratory variables routinely used to assess the level of systemic inflammation in patients with RA are the ESR and CRP and, less frequently, platelet counts and fibrinogen (17). All of these four variables are ascribable to the same theme in our analysis that we therefore labeled ”acute phase response“. Anemia of chronic disease usually develops in an
12
environment of systemic inflammation, i.e. the same context as the acute phase response. Despite being perceived in this context, hemoglobin is not defined as a variable of acute phase response sensu stricto. Both hemoglobin and variables of iron metabolism (iron – less reliable due to its fluctuations; ferritin; transferrin; and transferrin saturation) have been a part of the extended laboratory evaluation of patients with chronic inflammatory conditions such as RA (18). Indeed, ferritin and transferrin act as acute phase response proteins (19, 20), but that role is probably less prominent compared to their role as markers of iron metabolism: since ferritin and transferrin saturation were ascribable to a single theme at baseline - different from the “acute phase response” theme - we labeled that theme “iron metabolism”. Recognition of hepcidin as a link between systemic inflammation and reduction of iron availability has filled an important gap in the understanding of the pathogenesis of anemia of chronic disease including anemia of chronic inflammation in RA (21). However, the role of hepcidin exceeds the context of inflammation, since it is induced in hepatocytes in the state of iron overload, regulating the amount of total body iron (22). In the context of iron overload, the function of hepcidin is independent of IL-6, a key cytokine involved in inflammation and activation of the acute phase response. Therefore, from this mechanistic perspective, hepcidin can be perceived as acting at a crossroad between systemic inflammation and iron homeostasis (21, 23). Previous studies revealed a positive crosssectional association of hepcidin with markers of iron metabolism (primarily ferritin) and acute phase reactants such as CRP, ESR and haptoglobin (7, 8, 21, 24). Indeed, in our study, a dual role of hepcidin was observed using PCA, revealing high loadings on the “iron metabolism” theme at a steady state, as well as on the “acute phase response” theme for its change scores after initiation of therapy. Not only does the hepcidin level increase in the 13
setting of inflammation, which is an action driven by IL-6 as the necessary and sufficient cytokine (25), but it also rapidly decreases (within days) if IL-6 activity goes down. The latter effect was observed with the administration of tocilizumab, an IL-6 receptor blocker (7-8). IL-6 receptor blockade seems to be superior to tumor necrosis factor alpha (TNF-α) inhibition in lowering hepcidin and improving hemoglobin levels (8). This finding may support the role of IL-6 as a pivotal cytokine driving anemia of RA. Despite the associations observed in the PCA setting, baseline hepcidin and its 12-week changes were not correlated with hemoglobin in univariable analysis. In the multivariable models, baseline hemoglobin was associated with variables of iron metabolism other than hepcidin (Figure 2, left panel). A lacking direct association of hepcidin and hemoglobin has been reported both cross-sectionally and longitudinally (7, 24). Knowing that hepcidin plays a role in the pathogenesis of anemia of chronic disease/inflammation, failure to identify an independent association between hemoglobin and hepcidin could be explained by a more complex link between inflammation and anemia where the hepcidin-mediated mechanism may not be dominant, or at least not direct (7). There is a growing knowledge of the potential effect of pro-inflammatory cytokines, such as TNF-α, IL-1 and interferon-γ on bone marrow suppression (26, 27), in vitro and in mouse models. Furthermore, TNF-α has been associated with the apoptotic depletion of hematopoietic precursor cells in patients with RA (28), illustrating its direct effect on the development of anemia, bypassing the previously described mechanism involving hepcidin. Nonetheless, hematopoietic precursor cell ex vivo cultures did not fill the gap of unexplained hemoglobin variability in either of the two models assessed in our study:
14
neither RA disease activity nor the extent of anemia was reflected by changes in the progenitor cell levels. Our study has several limitations. PCA is used for exploratory purposes. Loadings by PCA do not prove any biological association. The methods used in our study served the purpose to increase our ability to interpret the meaning and place of hemoglobin in the complex network of variables used to describe inflammation and anemia of RA. For this reason we did not include healthy individuals in this study. A further limitation may be the short followup period, which certainly allowed us to investigate the associations of change scores, but which was at the same time too short to assess long-term effects on harder outcomes, such as radiographic progression. Also, although we are confident that we have covered all possible domains in relation to anemia (and its causes), we were not able to assess the impact of cytokines or to adjust for comorbidities.
5. CONCLUSIONS In summary, this analysis allowed us to obtain an unprecedented, multi-faceted, crosssectional and short-term longitudinal view of the meaning of hemoglobin in RA. In addition to its dependence on body iron status, it seems that hemoglobin may act as a disease activity measure, being at least as informative as “classical” variables associated with the acute phase response, if not more informative. In the light of all these associations, a large proportion of the variability of hemoglobin values still remains unexplainable. Despite the well-described regulatory role of hepcidin, its association with anemia in RA is not directly apparent from our analyses, and remains to be fully understood.
15
Further studies may be warranted to assess the value of using serum hemoglobin levels to interpret a patient’s inflammatory status and even to guide treatment decisions, especially when compared to the role of parameters used in the standard assessment of disease activity in RA.
CONFLICTS OF INTEREST: All authors declare that they have no conflict of interest. Thasia Woodworth was previously the Tocilizumab Clinical Science leader for Roche and contributed to study design and conduct. Roche performed the measurement of hepcidin. This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.
REFERENCES 1. Weiss G, Goodnough LT. Anemia of chronic disease. N Engl J Med 2005; 352: 1011–23. 2. Smolen JS, Aletaha D. Scores for all seasons: SDAI and CDAI. Clin Exp Rheumatol 2014; 32(5 Suppl 85): S–75–9. 3. van der Heijde DM, van’t Hof MA, van Riel PL, van Leeuwen MA, van Rijswijk MH, van de Putte LB. Validity of single variables and composite indices for measuring disease activity in rheumatoid arthritis. Ann Rheum Dis 1992; 51: 177–81. 4. Han C, Rahman MU, Doyle MK et al. Association of anemia and physical disability among patients with rheumatoid arthritis. J Rheumatol 2007; 34: 2177–82.
16
5. Möller B, Scherer A, Förger F, Villiger PM, Finckh A; Swiss Clinical Quality Management Program for Rheumatic Diseases. Anaemia may add information to standardised disease activity assessment to predict radiographic damage in rheumatoid arthritis: a prospective cohort study. Ann Rheum Dis 2014; 73: 691–6. 6. van Steenbergen HW, van Nies JA, van der Helm-van Mil AH. Anaemia to predict radiographic progression in rheumatoid arthritis. Ann Rheum Dis 2013; 72: e16. 7. Isaacs JD, Harari O, Kobold U, Lee JS, Bernasconi C. Effect of tocilizumab on haematological markers implicates interleukin-6 signalling in the anaemia of rheumatoid arthritis. Arthritis Res Ther 2013; 15: R204. 8. Song SN, Iwahashi M, Tomosugi N et al. Comparative evaluation of the effects of treatment with tocilizumab and TNF-α inhibitors on serum hepcidin, anemia response and disease activity in rheumatoid arthritis patients. Arthritis Res Ther 2013; 15: R141. 9. Hashimoto M, Fujii T, Hamaguchi M et al. Increase of hemoglobin levels by anti-IL-6 receptor antibody (tocilizumab) in rheumatoid arthritis. PloS One 2014; 9: e98202. 10. Papadaki HA, Kritikos HD, Valatas V, Boumpas DT, Eliopoulos GD. Anemia of chronic disease in rheumatoid arthritis is associated with increased apoptosis of bone marrow erythroid cells: improvement following anti-tumor necrosis factor-alpha antibody therapy. Blood 2002; 100: 474–82. 11. Aletaha D, Neogi T, Silman AJ et al. 2010 Rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Arthritis Rheum 2010; 62: 2569–81.
17
12. Kobold U, Dülffer T, Dangl M et al. Quantification of hepcidin-25 in human serum by isotope dilution micro-HPLC-tandem mass spectrometry. Clin Chem 2008; 54: 1584-6. 13. Raj DS. Role of interleukin-6 in the anemia of chronic disease. Semin Arthritis Rheum 2009; 38: 382–8. 14. Sagaster V, Jäger E, Weltermann A et al. Circulating hematopoietic progenitor cells predict survival in patients with myelofibrosis with myeloid metaplasia. Haematologica 2003; 88: 1204–12. 15. Field A. Exploratory factor analysis. In Field A: Discovering statistics using SPSS. 2nd ed., Los Angeles-London-New Delhi-Singapore-Washington DC, Sage 2009: 627–85. 16. Kanfer EJ, Nicol BA. Haemoglobin concentration and erythrocyte sedimentation rate in primary care patients. J R Soc Med 1997; 90: 16–8. 17. Scott DL, Wolfe F, Huizinga TW. Rheumatoid arthritis. Lancet 2010; 376: 1094–108. 18. Osterholm EA, Georgieff MK. Chronic Inflammation and Iron Metabolism. J Pediatr 2015; 166: 1351–7.e1. 19. Gruys E, Toussaint MJM, Niewold TA, Koopmans SJ. Acute phase reaction and acute phase proteins. J Zhejiang Univ Sci 2005; 6B: 1045–56. 20. Yildirim K, Karatay S, Melikoglu MA, Gureser G, Ugur M, Senel K. Associations between acute phase reactant levels and disease activity score (DAS28) in patients with rheumatoid arthritis. Ann Clin Lab Sci 2004; 34: 423–6.
18
21. Demirag MD, Haznedaroglu S, Sancak B et al. Circulating hepcidin in the crossroads of anemia and inflammation associated with rheumatoid arthritis. Intern Med 2009; 48: 421–6. 22. Singh B, Arora S, Agrawal P, Gupta SK. Hepcidin: a novel peptide hormone regulating iron metabolism. Clin Chim Acta 2011; 412: 823–30. 23. Choy E. Understanding the dynamics: pathways involved in the pathogenesis of rheumatoid arthritis. Rheumatol 2012; 51 Suppl 5: v3–11. 24. Sellam J, Kotti S, Fellahi S et al. Serum hepcidin level is not an independent surrogate biomarker of disease activity or of radiographic progression in rheumatoid arthritis: results from the ESPOIR cohort. Ann Rheum Dis 2013; 72: 312–4. 25. Nemeth E, Rivera S, Gabavan V et al. IL-6 mediates hypoferremia of inflammation by inducing the synthesis of the iron regulatory hormone hepcidin. J Clin Invest 2004; 113: 1271-6. 26. Libregts SF, Gutiérrez L, de Bruin AM et al. Chronic IFN-γ production in mice induces anemia by reducing erythrocyte life span and inhibiting erythropoiesis through an IRF-1/PU.1 axis. Blood 2011; 118: 2578–88. 27. Nemeth E, Ganz T. Anemia of inflammation. Hematol Oncol Clin North Am 2014; 28: 671– 81, vi. 28. Papadaki HA, Kritikos HD, Gemetzi C et al. Bone marrow progenitor cell reserve and function and stromal cell function are defective in rheumatoid arthritis: evidence for a tumor necrosis factor alpha-mediated effect. Blood 2002; 99: 1610–9.
19
TABLES Table 1. Clinical measures and laboratory variables measured at baseline and week 12. Parameter Baseline alla Baseline activeb Week 12 activec CDAI 6.9 (0.85-15.48) 17.5 (12.8-24.8) 12.3 (6.8-22.6) SDAI 7 (1.35-16.57) 20.7 (13.88-26.48) 13.58 (7.37-24) DAS28-ESR 2.97 (2.1-4.01) 4.53 (3.68-5.39) 3.49 (2.58-4.92) SJC28 1 (0-4) 4 (3-7) 4 (2-6) TJC28 0 (0-2) 4 (2-10) 2 (0-5) PGA 23 (3.25-49.75) 50 (39-64) 40 (14-50) EGA 12 (0-25.75) 28 (22-37) 20.5 (13-33.5) Pain (VAS) 20 (3-43.75) 44 (31-63) 32 (17-45) HAQ 0.25 (0-1) 1 (0.38-1.38) 0.63 (0.38-1) MST (minutes) 0 (0-27.5) 30 (0-60) 10 (0-52.6) RF (U/L) 25 (12-129) 51.05 (12-209.25) 33 (12-169) ACPA (U/L) 103 (2.1-600) 154 (1.6-600) 132 (1.5-600) ESR (mm/h) 16 (9-25) 17 (9.75-26) 15 (8-22.5) CRP (mg/dL) 0.44 (0.17-0.98) 0.58 (0.25-2.24) 0.39 (0.19-0.91) Fibrinogen (mg/dL) 397 (340.5-482.25) 408 (357.75-487.25) 381 (349-459) Hepcidin (nmol/L) 4.65 (1.03-12.43) 6.72 (1.05-14.75) 3.21 (0.51-10.81) Erythrocytes (x10E12/L) 4.4 (4.2-4.7) 4.4 (4.1-4.7) 4.5 (4.2-4.8) Hemoglobin (g/dL) 13.2 (12.4-14.1) 12.8 (11.8-14.1) 13.1 (12.2-13.9) Hematocrite (%) 39.8 (37.6-42.1) 39.3 (36-41.8) 39.5 (36.7-41.6) MCV (fL) 90.2 (85.8-93.3) 88.9 (82.3-93.2) 89 (84.4-92.2) Reticulocytes (x10E9/L) 41.3 (33.7-59.05) 42.7 (30.9-65.2) 45.55 (35.6-62.73) Leukocytes (x10E9/L) 7.78 (6.11-9.85) 8.5 (6.36-11.1) 7.64 (5.59-11.08) Platelets (x10E9/L) 260 (222-295) 270 (236-312) 259 (226-289) Iron (ug/dL) 71 (51-92.75) 64 (37-89) 61 (41-99) Transferrin (ug/dL) 257.65 (232.83-287.3) 256.2 (227.6-287.3) 260.5 (240.6-293.2) Transferrin saturation (%) 19.8 (13.38-27.4) 18.3 (11.3-25.7) 18 (10.3-31.4) Ferritin (ng/mL) 55.95 (29.8-94.88) 52.4 (30.2-91.2) 38.8 (25.5-92.8) BFU-E (ml-1) 641 (410-1004) 636 (481-916) 626.5 (325.5-968) -1 CFU-GM (ml ) 86 (53.25-139.75) 91 (65-132) 70 (36.5-145.25) -1 CFU-GEMM (ml ) 11 (0-22) 11 (3-21) 12 (4-15.25) a values at beaseline for all patients (left); bvalues at baseline for initially active patients (centre); cvalues at week 12 for initially active patients (right); *baseline active and week 12 active compared (Wilcoxon signed ranks test (2-tailed)); data are presented as medians and interquartile ranges.
20
p* 0.009 0.006 0.004 0.415 0.003 0.003 0.013 0.002 0.212 0.058 0.007 0.147 0.024 0.057 0.158 0.083 0.162 0.174 0.993 0.222 0.517 0.497 0.064 0.478 0.062 0.774 0.092 0.778 0.200 0.333
Table 2. Principal component analysis – loadings of variables to each theme. Loadings presented at baseline and week 12 (for all patients), as well as at changes between baseline and week 12 (for initially active patients)
Included
Signatu re* for
CDAI
(1)
SDAI DAS28ESR
(1) (1)
SJC28 TJC28 EGA PGA HAQ Pain (VAS) MST RF CCP ESR
(3)
CRP Fibrinoge n Erythrocy tes Hemoglo bin Hematocr it
(3)
MCV Reticuloc ytes Leukocyt
(3)
(1) Clinical (2) (4) Iron activity Hematopoiesis (3) Acute phase metabolism BL 12 Δ BL 12 Δ BL 12 Δ BL 12 0.9 0.9 1.0 0.0 0.0 0.0 0.0 0.0 0.1 0.1 0.0 68 98 00 71 29 01 04 91 31 05 00 0.9 0.9 0.9 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 36 64 10 85 38 28 10 58 96 71 09 0.8 0.8 0.8 0.1 0.1 0.0 0.2 0.1 0.0 0.0 0.0 01 42 54 90 71 25 07 89 46 26 69 0.7 0.6 0.4 0.2 0.0 0.0 0.1 0.0 0.4 0.0 0.2 42 74 51 04 80 38 44 53 06 97 65 0.7 0.7 0.9 0.0 0.0 0.0 0.0 0.2 0.3 0.1 0.2 46 86 39 07 32 55 03 00 27 99 38 0.8 0.8 0.9 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.1 85 48 00 69 32 96 84 52 13 94 72 0.7 0.8 0.3 0.0 0.0 0.1 0.2 0.0 0.1 0.1 0.0 88 42 25 15 53 66 23 74 70 03 69 0.4 0.5 0.1 0.0 0.1 0.1 0.1 0.0 0.1 0.0 0.3 18 61 54 89 37 60 25 74 06 65 01 0.7 0.7 0.1 0.0 0.0 0.2 0.2 0.0 0.3 0.0 0.0 85 85 29 52 99 06 09 12 88 97 29 0.6 0.5 0.2 0.0 0.1 0.1 0.1 0.1 0.2 0.1 0.0 23 27 23 43 87 84 98 19 34 80 66 0.1 0.0 0.0 0.2 0.1 0.0 0.0 0.1 0.0 0.1 0.0 17 84 63 27 58 57 18 37 62 35 69 0.1 0.2 0.1 0.3 0.2 0.0 0.0 0.1 0.0 0.0 0.0 62 81 91 11 05 86 64 74 68 53 82 0.1 0.0 0.1 0.4 0.1 0.1 0.5 0.8 0.5 0.2 0.0 01 56 03 36 78 54 58 02 12 20 14 0.1 0.0 0.0 0.1 0.1 0.0 0.7 0.6 0.9 0.1 0.0 20 38 06 22 20 28 19 92 08 93 47 0.0 0.1 0.4 0.1 0.1 0.0 0.6 0.6 0.5 0.4 0.1 83 70 58 19 59 70 46 40 64 29 95 0.0 0.0 0.2 0.6 0.7 0.1 0.0 0.0 0.4 0.4 0.0 36 22 08 55 43 78 89 20 55 02 43 0.1 0.1 0.6 0.4 0.6 0.1 0.3 0.2 0.1 0.6 0.5 39 17 88 57 46 51 96 53 69 47 41 0.0 0.0 0.6 0.5 0.6 0.1 0.3 0.2 0.3 0.6 0.4 96 82 07 17 58 57 59 29 11 17 54 0.2 0.0 0.0 0.1 0.1 0.2 0.4 0.2 0.6 0.0 0.6 34 82 28 18 74 47 46 81 71 21 57 0.2 0.0 0.1 0.0 0.4 0.6 0.0 0.0 0.1 0.1 0.1 70 75 50 95 35 72 31 47 11 21 20 0.1 0.0 0.0 0.3 0.2 0.4 0.4 0.6 0.0 0.0 0.0
Δ 0.0 25 0.1 03 0.2 09 0.3 08 0.0 07 0.1 33 0.3 10 0.5 33 0.3 92 0.5 01 0.7 34 0.3 45 0.0 78 0.0 32 0.1 53 0.0 56 0.0 55 0.0 33 0.0 45 0.1 36 0.2 21
es Platelets Iron Transferri n Transf. Sat.
(4)
Ferritin
(4)
(4) (4)
Hepcidin BFU-E
(2)
CFU-GM CFUGEMM
(2) (2)
31 0.0 11 0.0 77 0.0 64 0.0 65 0.0 23 0.1 84 0.0 30 0.0 37 0.0 17
44 0.0 43 0.0 84 0.0 12 0.0 76 0.0 03 0.1 54 0.0 51 0.0 52 0.0 16
28 0.0 99 0.0 58 0.1 52 0.0 90 0.2 56 0.0 56 0.0 19 0.0 07 0.0 91
90 0.0 27 0.1 94 0.2 53 0.2 63 0.0 67 0.1 12 0.8 24 0.7 84 0.7 53
67 0.0 06 0.0 95 0.0 75 0.0 93 0.2 69 0.1 04 0.8 25 0.7 64 0.6 66
25 0.4 66 0.0 16 0.4 62 0.1 10 0.3 33 0.0 72 0.8 11 0.8 44 0.8 60
67 0.5 04 0.6 74 0.0 43 0.6 60 0.0 69 0.0 14 0.0 75 0.1 07 0.0 90
24 0.7 28 0.5 43 0.2 37 0.5 24 0.1 41 0.4 54 0.1 41 0.0 92 0.0 97
01 0.2 68 0.4 43 0.5 39 0.4 00 0.5 04 0.5 79 0.0 73 0.1 65 0.1 17
89 0.1 24 0.2 43 0.4 63 0.3 06 0.7 26 0.4 99 0.0 12 0.1 38 0.0 41
39 0.0 93 0.6 13 0.4 92 0.6 26 0.6 19 0.5 91 0.0 55 0.1 53 0.1 42
65 0.0 37 0.6 66 0.0 80 0.7 13 0.4 58 0.0 59 0.1 64 0.0 72 0.1 39
bold – loadings over 0.400; BL – baseline; 12 – week 12; Δ – changes between baseline and week 12; Transf. Sat. – transferrin saturation; *the variables with the highest loading leading to the labelling (“theme”) of a factor are indicated as signature variables for that respective factor.
22
FIGURES
Figure 1. Loadings of hemoglobin on the four themes. Full line: loadings at baseline (all patients); dotted line: loadings for changes between baseline and week 12 (patients active at baseline)
(note to the Editor: single-column fitting image)
23
Figure 2. Percentage of variability of hemoglobin explained by individual variables in two multivariable linear models. Baseline: model explaining the variability of baseline hemoglobin (all patients). Change scores: model explaining the variability of hemoglobin changes between baseline and week 12 (patients active at baseline). EGA – evaluator global assessment; ESR – erythrocyte sedimentation rate.
(note to the Editor: single-column fitting image)
24