Background: The closure of long-standing gaps in our knowledge of aetiological factors behind anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) is a major challenge. Descriptive and analytical epidemiological studies can improve our understanding of environmental inﬂuences. Reported seasonal variations in AAV, mainly related to Wegener’s disease, have shown an increasing number of cases in the winter months, which could be related to an extrinsic factor underlying infection. The objective of this paper was to study seasonal variations in AAV with respect to renal affectation diagnosed in Catalonia, Spain. Methods: Two hundred and thirty-four patients diagnosed for renal AAV between 2001 and 2014 in eight hospitals in Catalonia were included in the study. We used medical records to retrospectively analyse the date of the ﬁrst symptoms attributed to the AAV, ANCA subtypes, the degree of renal impairment and renal histology. Results: Of the 234 patients studied, 49.2% were male and 50.8% female. For ANCA status, 8.5% were positive, 15.9% were proteinase-3-positive and 75.6% were myeloperoxidase-positive. In relation to histological classiﬁcation, 17.8% were sclerotic, 11.7% focal, 38.8% crescentic and 31.7% mixed. Regarding seasonal distribution, we observed a clear seasonal periodicity with a signiﬁcantly higher incidence of cases in the winter. Applying an Eigen decomposition, we observed a periodic ﬂuctuation of frequencies around the annual cycle with peaks every 10–12 months, and higher incidence of AAV cases in February. Conclusions: Our results conﬁrm, in Catalonia, the seasonal periodicity of AAV with a higher incidence in the winter, as formerly described in the literature for other regions. An environmental factor, likely one that is infectious, may explain this ﬁnding. Key words: ANCA, epidemiology, immunology, seasonal, seasonality, vasculitis Received: June 12, 2017. Editorial decision: September 25, 2017 V C The Author 2017. Published by Oxford University Press on behalf of ERA-EDTA. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/ licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact email@example.com Downloaded from https://academic.oup.com/ckj/article-abstract/11/4/468/4683650 by Ed 'DeepDyve' Gillespie user on 07 August 2018 Seasonal variation in ANCA-associated vasculitis | 469 subtype, the degree of renal impairment and renal histology classi- Introduction fication. Renal pathology was classified according to the Berden Anti-neutrophil cytoplasmic antibody (ANCA)-associated classification as follows: focal, crescentic, mixed or sclerotic . vasculitis (AAV) comprises three different clinical entities— Related to the date of the first AAV symptoms, we included granulomatosis with polyangiitis (GPA), microscopic polyangiitis onset data for general symptoms such as fever, malaise and/or (MPA) and eosinophilic granulomatosis with polyangiitis—a group weight loss or specific organ involvement, and ear, nose and of systemic diseases that are characterized by pauci-immune nec- throat, pulmonary, renal, ophthalmological and cutaneous in- rotizing inflammation of small to medium vessels. Present in volvement. The disease onset–diagnosis interval was calculated most patients, ANCAs against proteinase-3 (PR3) or myeloperoxi- as the difference between the onset of symptoms and the initia- dase (MPO) are often associated with GPA and MPA, respectively tion of AAV treatment. For these calculations, onset data were . AAV affects multiple organ systems, with glomerulonephritis arbitrarily set as the 15th day of the respective month, unless and pulmonary involvement being frequent manifestations. AAV patients were able to specify the exact week or day of disease is currently defined as a rare autoimmune disease with a current onset. Finally, we excluded 11 patients because a precise month annual incidence of 20/million/year , with increasing incidence of the onset of AAV could not be calculated. in recent years . The incidence of PR3-AAV and MPO-AAV varies Data analysis was performed using GraphPad Prism 6.0 worldwide. In general, MPO-AAV is seen more frequently in south- (GraphPad Software, San Diego, CA, USA). The distribution of ern Europe, Asia and the Pacific (except in New Zealand and symptom onset according to month and season was examined Australia), while PR3-AAV is more common in the northern coun- for uniformity using exact one-way goodness-of-fit chi-square tries of the world [4–7]. In Spain, a study of 450 patients with AAV tests as a means to identify significant deviations from ex- found the proportions of entities to be 40.9% for GPA, 37.1% for pected frequencies. Seasons were divided into spring (April– MPA and 22% for eosinophilic granulomatosis with polyangiitis June), summer (July–September), autumn (October–December) . In another study, which analysed 151 patients with AAV and and winter (January–March). renal involvement, proportions of 39.7% for MPA, 6.6% for GPA An Eigen decomposition was applied to the ANCA time se- and 53.6% for renal limited vasculitis were found . ries with the covariance matrix equivalent of processing a for- The aetiology of AAV is currently unknown, and so too are the ward–backward prediction data matrix by signal strength rather contributions of genetic and environmental factors to its develop- than by frequency. Due to the low signal-to-noise ratio in the ep- ment. Over the past few years, genetic factors contributing to idemiological time series, it was possible in this way to isolate AAV have been extensively studied, mostly by genome-wide as- individual oscillatory components embedded in signals. In sociation studies. These have demonstrated that patients with this decomposition, the first eigenvalue, corresponding to sea- PR3-ANCA have significant association with human leukocyte sonality, accounts for 6.5% of the normalized variability. To fur- antigen (HLA)-DP, and with genes encoding a1-antitrypsin (Serpin ther cross-validate this result, the series of cases was Family A Member 1 (SERPINA1), the endogenous inhibitor of PR3) accumulated and then detrended by a linear least squares and PR3 (proteinase 3; PRTN3). On the other hand, patients with approximation. MPO-ANCA have significant association with HLA-DQ . Comparisons of seasonal distribution patterns for patient In relation to environmental factors, previous studies have groups (sex, ANCA subtype, degree of renal impairment and re- demonstrated an increased incidence of AAV in patients exposed nal histology classification) were performed using a chi-square to a variety of air pollutants (e.g. silica) , after treatment with test. Regarding renal impairment, we divided patients into two drugs such as penicillamine and hydralazine [10, 11], and in rela- different groups based in their baseline serum creatinine: group tion to infectious organisms like a nasal colonization with 1, <500 lmol/L; and group 2, 500 lmol/L. Regarding histology Staphylococcus aureus in GPA patients . Supporting the idea of classification, we also divided the patients into two different an underlying infectious factor, several studies have shown that groups: the first one included patients with crescentic, focal and the onset of AAV varies by season, with incidence peaking in the mixed classification; and the second one included only patients winter [13–17]. In clear contrast, a recent study suggested that with sclerotic classification. This decision was based on the fact AAV appears preferentially in the summer in GPA patients , that patients with sclerotic lesions usually have a more impre- supporting the idea of a possible allergic mechanism in its patho- cise date of first symptoms of disease. genesis. In addition, in other primary systemic vasculitis condi- Results were considered significant when P< 0.05 and data tions, such as Kawasaki disease, a seasonal pattern and possible are expressed as mean 6 standard error of the mean. environmental triggers have been shown [19, 20]. Written informed consent was considered not to be neces- In the present study, we re-examined the hypothesis of sea- sary for the study, as it was a retrospective analysis of our usual sonal variations in the onset of renal AAV in a Mediterranean everyday work. The data of the patients were anonymized for area in Spain. the purposes of this analysis. The confidential information of the patients was protected according to national norms. This manuscript has been approved for publication by the Clinical Materials and methods Research Ethics Committee of Bellvitge University Hospital. This retrospective study included 234 patients diagnosed with AAV with renal involvement between January 2001 and December 2014 in eight different hospitals in Catalonia, Spain. Diagnosis of renal Results vasculitis was made by according to the criteria established at the Characteristics of the study population Chapel Hill Conference , as determined by positive ANCA (MPO or PR3) antibodies and a renal biopsy with the presence of necrotiz- The main characteristics of our patients at AAV diagnosis are ing pauci-immune glomerulonephritis. summarized in Table 1. Most of the 234 patients (around 88%) Information regarding the following demographics were ob- lived in the city of Barcelona or in its metropolitan area, with a tained from medical records: age, gender, disease features, the date predominance of an urban and Caucasian population. of first symptoms attributed to the AAV, date of diagnosis, ANCA Regarding the demographic characteristics, 49.2% were male Downloaded from https://academic.oup.com/ckj/article-abstract/11/4/468/4683650 by Ed 'DeepDyve' Gillespie user on 07 August 2018 470 | J. Draibe et al. and 50.8% female; 15.9% were PR3-ANCA-positive, 75.6% MPO- Seasonal variations of AAV onset related to geographic ANCA-positive and 8.5% ANCA-negative. For renal impairment, location the mean baseline serum creatinine at the diagnosis was 422.93 In our study, we included patients from eight different hospitals lmol/L (range 62–1418). In respect of histological classification, and areas were as follows: one was urban, located in the city of 17.8% were sclerotic, 11.7% focal, 38.8% crescentic and 31.7% Barcelona (Area 1, n ¼ 97); two were located in the metropolitan mixed. The mean onset–diagnosis interval calculated was 3.15 area of Barcelona (Area 2, n¼ 70 and Area 3, n¼ 39; located west months (range 0.1–45.5). and east of Barcelona, respectively); two were located northeast of Catalonia (Area 4, n ¼ 22); and one was located north-west of Catalonia (Area 5, n¼ 6). Monthly and seasonal variations of AAV onset When we analysed the seasonal variation of AAV related to Monthly and seasonal distributions of AAV onset are shown in geographic location, we found that patients from Barcelona and Figure 1. Frequency of disease onset according to month the western metropolitan area had similar patterns, with more showed a higher incidence in January, February and March, but homogenous distribution across the seasons. On the other hand, there were no significant variations. patients from the east or north-east of Barcelona had a higher in- Regarding seasonal distribution, we observed a significantly cidence of AAV in the winter, whereas patients from the north- higher incidence of cases in the winter and this variation was west had higher incidence in spring and summer (Figure 3). significant when compared with the seasonal distribution re- ported in ANCA MPO-positive patients (chi-square ¼ 13.36, 3 de- grees of freedom, P¼ 0.003) . Discussion Applying an Eigen decomposition, we observed periodic fluc- AAV needs to be understood as a multifactorial disease that re- tuation of the frequencies with peaks every 10–12 months sults from environmental triggers affecting genetically around the annual cycle, and a greater incidence of AAV cases predisposed individuals. Investigating the effects of seasonal in February (Figure 2). climate and geographical variations on the onset of this disease Using stratification of selected variables (sex, ANCA subtype, could provide indirect information about its pathogenesis. degree of renal impairment and renal histology classification), Our study re-examined the hypothesis of seasonal variation no significant differences were found in the analyses of sea- in the onset of AAV in a Mediterranean area in north-east Spain sonal patterns (Table 2). and demonstrated higher incidence of AAV patients with renal affectation during the winter, supporting several previous stud- ies that have found similar results [13–17]. This could be linked Table 1. AAV patient characteristics to a higher incidence of airway disease in this season, and sub- sequent respiratory infection factors may trigger the disease, as Variables Baseline characteristics described previously ; however, our study could not provide Age, mean (SD), years 65.33 (16–89) support this proposed link. Sex (M/F), % 49.2/50.8 In addition, we demonstrated a periodic seasonal fluctuation Baseline creatinine, mean (range), lmol/L 422.93 (51–1418) related to the incidence of AAV, with peaks every 10–12 months <500 lmol/L, % 67.5 around the annual cycle. Tidman et al. previously described >500 lmol/L, % 32.5 periodic fluctuations in patients with AAV over a period of 21 ANCA subtype years; however, these peaks occurred around every 3–4 years. Negative, n (%) 20 (8.5) This difference could be related to the shorter period of our Anti-MPO, n (%) 177 (75.6) study compared with that of Tidman et al. Anti-PR3, n (%) 37 (15.9) Of note, most of the previous studies were performed in pop- Histology classiﬁcation ulations that included mostly GPA- or PR3-ANCA-positive pa- Focal, n (%) 26 (11.8) Crescentic, n (%) 85 (38.6) tients that do not match with our cohort, which was Mixed, n (%) 70 (31.8) predominantly composed of MPO-ANCA-positive patients. Only Sclerotic, n (%) 39 (17.8) one study analysed seasonal variations in a population of MPO- ANCA patients  and found that very few of them presented Fig. 1. Distribution of months (A) and season (B) of AAV onset for 234 patients. Downloaded from https://academic.oup.com/ckj/article-abstract/11/4/468/4683650 by Ed 'DeepDyve' Gillespie user on 07 August 2018 Seasonal variation in ANCA-associated vasculitis | 471 Fig. 2. Evolution of the number of ANCA cases (black) from 2001–14 together with their reconstructed seasonality (red). An Eigen decomposition of order M¼ 30 was ap- plied to the raw ANCA time series to maximize the signal extraction process. The inset shows the maximum entropy spectrum derived for this component, showing two clear signiﬁcant periods around the annual cycle (11.76 months and 10.12 months, P< 0.05). Fig. 3. Seasonal distribution of AAV patients in different hospitals included in the study. (A) Area 1 (Barcelona), Area 2 (L’Hospitalet de Llobregat), Area 3 (Badalona), Area 4 (Girona and Palamo ´ s) and Area 5 (Manresa). (B) The graph shows that patients from Barcelona (Area 1) and the south metropolitan area (Area 2) had similar pat- terns, with more homogenous distribution across the seasons. On the other hand, patients from the north metropolitan area or the north-east (Areas 3 and 4) had a greater incidence of AAV in the winter, while patients from the north-west (Area 5) had a greater incidence in the spring and summer. with onset of the disease during the summer (14%), with a simi- Interestingly, when we divided patients according to their lar distribution across the other seasons. In our cohort, fewer geographic location, higher incidence of AAV was found in the patients presented with MPO-ANCA in the summer, but this as- winter for patients located in areas with similar characteristics sociation was not statistically significant. (i.e. the east and north-east of Catalonia). It is possible that an With regard to ANCA subtypes, we found lower incidence in environmental factor influences the higher incidence of the dis- the spring for patients who were ANCA-negative, with only one pa- ease in these areas in the winter, a subject that deserves further tient’s(5%)symptoms initiatingduringthisseason. Although investigation. ANCA positivity is noted in more than 90% of AAV patients, several Some limitations of this study have to be considered. Since studies have observed negative ANCA in 10–20% of cases [24–28]. this study was retrospective and data about the onset of the first Clinically, the majority of these cases are limited to kidney affecta- symptoms were obtained from medical records, we have to con- tion with a lower incidence of extra-renal symptoms. A functional sider the possibility of recall bias and the difficulty of accurately effect of ANCA epitopes or different antibodies  (such as anti- determining the moment that this subacute disease first be- moesin , anti-endothelial cells  and anti-human lysosome- comes apparent. Mahr et al. analysed this problem in their associated membrane protein-2 ) has been implicated in the study, comparing data obtained from medical records and tele- pathogenesis of this disease, but the exact mechanism remains phone interviews, and found an average discrepancy of about 2.6 unclear. The seasonal difference observed between ANCA-positive months. However, they still found similar results using data col- and -negative patients in our study supports the idea that these lected from medical records. As a result, we discuss here the sea- two entities may have different pathogenic triggers. sonal nature of symptom presentation rather than the exact Downloaded from https://academic.oup.com/ckj/article-abstract/11/4/468/4683650 by Ed 'DeepDyve' Gillespie user on 07 August 2018 472 | J. Draibe et al. Table 2. Seasonal variation of AAV onset stratiﬁed for selected variables (P values refer to comparison distribution across data) Variables Patients (n) P value Season of onset, n (%) Winter Spring Summer Autumn Sex 234 0.5 Male 116 39 (33.6) 34 (29.3) 19 (16.4) 24 (20.7) Female 118 43 (36.4) 25 (21.2) 20 (17) 30 (25.4) Serum creatinine 234 60 (38) 34 (21.5) 27 (17.1) 37 (23.4) 0.45 <500 lmol/L 158 22 (29) 22 (29) 15 (19.7) 17 (22.3) >500 lmol/L 76 ANCA subtype 234 Negative 20 8 (40) 1 (5) 5 (25) 6 (30) 0.35 PR3 37 16 (43.2) 10 (27) 5 (13.6) 6 (16.2) MPO 177 57 (32.2) 43 (24.3) 34 (19.2) 43 (24.3) Histology 220 18 (46.2) 9 (23) 6 (15.4) 6 (15.4) 0.5 Sclerotic 39 66 (36.5) 39 (21.5) 31 (17) 45 (25) Others 181 month of occurrence. The other principal limitation of this study 7. O’Donnell JL, Stevanovic VR, Frampton C et al. Wegener’s is that only patients with AAV and renal affectation were in- granulomatosis in New Zealand: evidence for a latitude- cluded in this study; despite the large number of patients, this dependent incidence gradient. Intern Med J 2007; 37: approach might not be fully representative of AAV in general. 242–246 8. Lyons PA, Rayner TF, Trivedi S et al. Genetically distinct sub- sets within ANCA-associated vasculitis. N Engl J Med 2012; Conclusion 367: 214–223 9. Hogan SL, Cooper GS, Savitz DA et al. Association of silica ex- In conclusion, our results in a cohort of AAV patients with renal posure with anti-neutrophil cytoplasmic autoantibody affectation in Spain confirm a periodic fluctuation and seasonal small-vessel vasculitis: a population-based, case-control variation of AAV, as described in the literature, with higher inci- study. Clin J Am Soc Nephrol 2007; 2: 290–299 dence in the winter. These results make it plausible that envi- 10. Mathieson PW, Peat DS, Short A et al. Coexistent membranous ronmental factors, most likely an infection trigger, act as nephropathy and ANCA-positive crescentic glomerulonephritis activators for the development of the disease. The seasonal dif- in association with penicillamine. NephrolDialTransplant 1996; ferences observed between ANCA-positive and -negative pa- 11: 863–866 tients warrant further careful examination. 11. Short AK, Lockwood CM. Antigen speciﬁcity in hydralazine associated ANCA positive systemic vasculitis. QJM 1995; 88: Acknowledgements 775–783 12. Stegeman CA, Tervaert JW, Sluiter WJ et al. Association of X.R. acknowledges the support from the Daniel Bravo chronic nasal carriage of Staphylococcus aureus and higher re- Foundation through the grant Windbiome. The rest of the lapse rates in Wegener granulomatosis. Ann Intern Med 1994; authors received no specific funding for this work. 120: 12–17 13. Falk RJ, Hogan S, Carey TS et al. Clinical course of anti- Conflict of interest statement neutrophil cytoplasmic autoantibody-associated glomerulo- nephritis and systemic vasculitis. The Glomerular Disease None declared. Collaborative Network. Ann Intern Med 1990; 113: 656–663 14. Raynauld JP, Bloch DA, Fries JF. Seasonal variation in the on- References set of Wegener’s granulomatosis, polyarteritis nodosa and giant cell arteritis. J Rheumatol 1993; 20: 1524–1526 1. Finkielman JD, Lee AS, Hummel AM et al. ANCA are detect- 15. Carruthers DM, Watts RA, Symmons DP et al. Wegener’s able in nearly all patients with active severe Wegener’s gran- granulomatosis–increased incidence or increased recogni- ulomatosis. Am J Med 2007; 120: 643.e9–e14 tion? Br J Rheumatol 1996; 35: 142–145 2. Scott DG, Watts RA. Epidemiology and clinical features of 16. Blockmans D, Bley T, Schmidt W. Imaging for large-vessel systemic vasculitis. Clin Exp Nephrol 2013; 17: 607–610 vasculitis. Curr Opin Rheumatol 2009; 21: 19–28 3. Knight A, Ekbom A, Brandt L et al. Increasing incidence of 17. Tidman M, Olander R, Svalander C et al. Patients hospitalized Wegener’s granulomatosis in Sweden, 1975-2001. J because of small vessel vasculitides with renal involvement Rheumatol 2006; 33: 2060–2063 in the period 1975-95: organ involvement, anti-neutrophil cy- 4. Ntatsaki E, Watts RA, Scott DG. Epidemiology of ANCA- toplasmic antibodies patterns, seasonal attack rates and ﬂuc- associated vasculitis. Rheum Dis Clin North Am 2010; 36: 447–461 tuation of annual frequencies. JIntern Med 1998; 244: 133–141 5. Gibson A, Stamp LK, Chapman PT et al. The epidemiology of 18. Mahr A, Artigues N, Coste J et al. Seasonal variations in onset of Wegener’s granulomatosis and microscopic polyangiitis in a Wegener’s granulomatosis: increased in summer? JRheumatol Southern Hemisphere region. Rheumatology 2006; 45: 624–628 2006; 33: 1615–1622 6. Ormerod AS, Cook MC. Epidemiology of primary systemic 19. Rodo ´ X, Ballester J, Cayan D et al. Association of Kawasaki vasculitis in the Australian Capital Territory and south- disease with tropospheric wind patterns. Sci Rep 2011; 1: 152 eastern New South Wales. Intern Med J 2008; 38: 816–823 Downloaded from https://academic.oup.com/ckj/article-abstract/11/4/468/4683650 by Ed 'DeepDyve' Gillespie user on 07 August 2018 Seasonal variation in ANCA-associated vasculitis | 473 26. Minz RW, Chhabra S, Joshi K et al. Renal histology in pauci- 20. Rodo ´ X, Curcoll R, Robinson M et al. Tropospheric winds from northeastern China carry the etiologic agent of Kawasaki immune rapidly progressive glomerulonephritis: 8-year ret- disease from its source to Japan. Proc Natl Acad Sci U S A 2014; rospective study. Indian J Pathol Microbiol 2012; 55: 28–32 111: 7952–7957 27. Eisenberger U, Fakhouri F, Vanhille P et al. ANCA-negative 21. Jennette JC, Falk RJ, Andrassy K et al. J Proposal of an interna- pauci-immune renal vasculitis: histology and outcome. tional consensus conference. Arthritis Rheum 1994; 37: Nephrol Dial Transplant 2005; 20: 1392–1399 187–192 28. Furuta S, Jayne DR. Antineutrophil cytoplasm antibody- 22. Berden AE, Ferrario F, Hagen EC et al. Histopathologic classiﬁcation associated vasculitis: recent developments. Kidney Int 2013; of ANCA-associated glomerulonephritis. JAmSoc Nephrol 2010; 21: 84: 244–249 1628–1636 29. Nagao T, Suzuki K, Utsunomiya K et al. Direct activation of 23. DeRemee RA, McDonald TJ, Weiland LH. Wegener’s granulo- glomerular endothelial cells by anti-moesin activity of anti- matosis: observations on treatment with antimicrobial myeloperoxidase antibody. Nephrol Dial Transplant 2011; 26: agents. Mayo Clin Proc 1985; 60: 27–32 2752–2760 24. Chen M, Yu F, Wang SX et al. Antineutrophil cytoplasmic 30. Cong M, Chen M, Zhang JJ et al. Anti-endothelial cell antibodies in antineutrophil cytoplasmic antibodies negative pauci-immune autoantibody-negative Pauci-immune crescentic glomerulo- nephritis. J Am Soc Nephrol 2007; 18: 599–605 crescentic glomerulonephritis. Nephrology 2008; 13: 228–234 25. Hung PH, Chiu YL, Lin WC et al. Poor renal outcome of anti- 31. Kain R, Tadema H, McKinney EF et al. High prevalence of neutrophil cytoplasmic antibody negative Pauci-immune autoantibodies to hLAMP-2 in anti-neutrophil cytoplasmic glomerulonephritis in Taiwanese. J Formos Med Assoc 2006; antibody-associated vasculitis. J Am Soc Nephrol 2012; 23: 105: 804–812 556–566 Downloaded from https://academic.oup.com/ckj/article-abstract/11/4/468/4683650 by Ed 'DeepDyve' Gillespie user on 07 August 2018
Clinical Kidney Journal – Oxford University Press
Published: Aug 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera