ET indicates essential thrombocythemia; MF, myelofibrosis; MPN-U, unclassifiable MPNs; and PV, polycythemia vera.
aFor some patients, fewer than 10 controls could be matched owing to old age (4 patients had 5 controls, 2 patients had 6 controls, 2 patients had 9 controls, and 8315 patients had 10 controls).
Years are since 30 days after index date.
eTable 1. Diagnostic Codes for MPNs, Age-Related Macular Degeneration, and Smoking-Related Diseases in the DNPR
eTable 2. Prevalence of Age-Related Macular Degeneration in Patients With MPN and Their Matched Controls
eTable 3. Number of Recorded Age-Related Macular Degeneration Events Among Patients With MPN in the DNPR
Customize your JAMA Network experience by selecting one or more topics from the list below.
Bak M, Sørensen TL, Flachs EM, et al. Age-Related Macular Degeneration in Patients With Chronic Myeloproliferative Neoplasms. JAMA Ophthalmol. 2017;135(8):835–843. doi:10.1001/jamaophthalmol.2017.2011
Do patients with chronic myeloproliferative neoplasms have an increased risk of age-related macular degeneration compared with the general population?
In a large Danish registry-based nationwide cohort study including 7958 patients with myeloproliferative neoplasms and 77 445 age- and sex-matched controls, the risk of age-related macular degeneration was increased for patients with myeloproliferative neoplasms, after adjustment for smoking and risk time.
These data suggest that patients with myeloproliferative neoplasms have a higher risk of age-related macular degeneration than the general population, supporting the possibility that systemic alterations may be involved in the pathogenesis of age-related macular degeneration.
It has been suggested that systemic inflammation increases the risk of age-related macular degeneration (AMD). Given that chronic immune modulation is present in patients with myeloproliferative neoplasms (MPNs), the risk of AMD in these patients may be increased.
To compare the risk of AMD in patients with MPNs with the risk of AMD in matched controls from the general population.
Design, Setting, and Participants
A nationwide population-based cohort study using Danish registers was conducted of all patients in Denmark who received a diagnosis between January 1, 1994, and December 31, 2013, of essential thrombocythemia, polycythemia vera, myelofibrosis, or unclassifiable MPNs. For each patient, 10 age- and sex-matched controls were included. All patients without prior AMD were followed up from the date of diagnosis (or corresponding entry date for the controls) until the first AMD diagnosis, death or emigration, or December 31, 2013, whichever occurred first. Data analysis was performed from April 1, 2015, to October 31, 2016.
Main Outcomes and Measures
Incidence of AMD recorded in specialized hospital-based care. The rates and absolute risk of AMD were calculated. Using Cox proportional hazards regression models, smoking and risk-time adjusted hazard ratios (HRs) between patients and controls were calculated. In addition, HRs of neovascular AMD after 2006 were calculated since antivascular endothelial growth factor treatment was introduced nationwide at hospitals thereafter.
A total of 7958 patients with MPNs (4279 women [53.8%] and 3679 men [46.2%]; mean [SD] age at diagnosis, 66.4 [14.3] years) were included in the study. The rate of AMD per 1000 person-years at risk was 5.2 (95% CI, 4.6-5.9) for patients with MPNs (2628 with essential thrombocythemia, 3063 with polycythemia vera, 547 with myelofibrosis, and 1720 with unclassifiable MPNs) and 4.3 (95% CI, 4.1-4.4) for the 77 445 controls, while the 10-year risk of AMD was 2.4% (95% CI, 2.1%-2.8%) for patients with MPNs and 2.3% (95% CI, 2.2%-2.4%) for the controls. The risk of AMD was increased overall for patients with MPNs (adjusted HR, 1.3; 95% CI, 1.1-1.5), with adjusted HRs for the subtypes of 1.2 (95% CI, 1.0-1.6) for essential thrombocythemia, 1.4 (95% CI, 1.2-1.7) for polycythemia vera, 1.7 (95% CI, 0.8-4.0) for myelofibrosis, and 1.5 (95% CI, 1.1-2.1) for unclassifiable MPNs. In addition, patients with MPNs had a higher risk of neovascular AMD (adjusted HR, 1.4; 95% CI, 1.2-1.6).
Conclusions and Relevance
Our results suggest that patients with MPNs are at increased risk of AMD, supporting the possibility that systemic inflammation is involved in the pathogenesis of AMD.
Age-related macular degeneration (AMD) is a chronic progressive eye disease that affects the macular region of the retina.1,2 The early and intermediate stages of the disease are characterized by drusen formation, with pigment changes also occurring in the intermediate stage. Patients may develop late-stage AMD, characterized by geographic atrophy or choroidal neovascularization (neovascular AMD), and some patients progress rapidly.3 Inflammation seems to be involved in the pathogenesis of AMD; recent studies have shown that systemic inflammation is also highly likely to be present.4 In addition, few studies have suggested an association between AMD and systemic diseases, in which chronic inflammation also plays a crucial role.5-7
Chronic myeloproliferative neoplasms (MPNs) are hematologic cancers characterized by clonal proliferation of 1 or more myeloid cell lineages. These neoplasms encompass essential thrombocythemia, polycythemia vera, myelofibrosis, and unclassifiable MPNs.8 Myeloproliferative neoplasms are heterogeneous but closely related diseases, with different cellular, molecular, and cytogenetic abnormalities as well as clinical presentations. However, these diseases have overlapping cellular and molecular alterations, and all MPNs are characterized by reduced survival among patients.9,10
The role of chronic systemic inflammation in patients with MPNs has recently been characterized;11 patients can experience symptoms and concurrent diseases driven by inflammation12,13 (eg, osteoporotic fractures,14 kidney diseases,15 and hematologic and nonhematologic cancers,16 as well as autoimmune phenomena17,18). Studies have revealed that some inflammatory pathways seen in patients with MPNs are similar to pathways demonstrated in patients with AMD. In particular, complement activation and altered nuclear factor erythroid 2–related transcription signaling appear to be involved in both diseases, potentially linking MPNs and AMD.17,19-28 We aimed to investigate the risk of AMD in patients with MPNs because this could have clinical relevance for these patients and, additionally, contribute to a better understanding of the inflammatory components in the pathogenesis of AMD.
In Denmark, patients with visual problems usually contact community-based general ophthalmologists; if signs of neovascular AMD are present, patients are then referred to hospital-based specialized care. Because the Danish health care system is a tax-funded public system with equal access for the entire population, the referral for and diagnosis of AMD are identical for everyone. In hospitals, the diagnosis is made by specialists in medical retinal diseases supported by multimodal imaging, including optical coherence tomography and fluorescein angiography (with registry-based record keeping of the diagnosis codes). National treatment guidelines for anti–vascular endothelial growth factor (anti-VEGF) suggest an initial loading dose with 3 monthly injections of anti-VEGF followed by an as-needed regimen.29 Because expenses pertaining to anti-VEGF treatment are reimbursed by government funds, neither the department nor the patient has any expenses related to treatment.
We undertook a matched cohort study using population-based nationwide registries. We could identify nationwide cohorts of patients with MPNs because physicians mandatorily assign and report diagnostic codes (according to the World Health Organization’s International Classification of Diseases, Eighth Revision, and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision) for patients in Danish hospitals to the Danish National Patient Registry (DNPR) nationwide.30 Initially, only diagnosis codes for inpatient admissions were reported, but since 1995, diagnosis codes for contacts at all emergency departments and outpatient hospital specialist clinics (including ophthalmology departments) have been reported. To compare the risk of AMD in (exposed) patients with MPNs with the risk in the general population, we included individually selected (unexposed) age- and sex-matched controls. This matching was feasible because individual-level linkage between different registries is possible via the Danish Civil Registration System.31 We obtained the follow-up status of the entire study population from the Civil Registration System. The registry includes information for each citizen on vital status (including date of death), emigration, and other demographics. All the diagnostic codes used in this study are shown in eTable 1 in the Supplement. This study was approved by the Danish Data Protection Agency. The use of Danish registry data did not require informed consent from the patients nor approval from the Danish Health Authority or the National Committee on Health Research Ethics.
All patients 18 years of age or older, registered with a first-ever inpatient or outpatient MPN diagnosis in the DNPR between January 1, 1994, and December 31, 2013, were included. Individuals with initial records of an MPN diagnosis but who subsequently had only recordings of “cytosis by other causes” (secondary polycythemia; code D75.1) were excluded before matching, which was done to avoid the inclusion of patients with incorrect MPN diagnoses. The index date was defined as the date of the first primary-listed diagnosis with essential thrombocythemia, polycythemia vera, myelofibrosis, or unclassifiable MPNs. For each patient, we selected 10 controls (unexposed to MPNs), matched by age (birth month and year) and sex. Controls had to be alive at the index date of their corresponding patient and were assigned with an identical index date. To ensure that only incident AMD events were used in the analysis, we excluded persons who had prior AMD. Because few patients with MPNs could present with blurred vision at diagnosis (caused by rheologic changes), coincidental diagnosis of asymptomatic AMD after ophthalmologic evaluation cannot be excluded. Thus, we also excluded events recorded until the first 30 days after the index date. Accordingly, we also excluded patients and controls who had less than 30 days of follow-up.
We used all incident AMD events recorded in the DNPR as the primary outcome measure. Because patients had been referred to highly specialized hospital departments for evaluation, we expect that most patients had neovascular AMD. However, because information on AMD staging is limited in the DNPR, the overall unspecific AMD code (H35.3) could be used regardless of stage (eg, for both intermediate-stage or late-stage dry AMD, as well as for neovascular AMD). To address this limitation, we therefore performed a subset analysis, estimating the risk using specific codes that were assigned only to persons who had neovascular AMD.
Follow-up began 30 days after the index date for all participants and continued until December 31, 2013, death, emigration, or an AMD diagnosis, whichever occurred first. Information on persons with AMD who received a diagnosis and/or were treated only by general ophthalmologists (and thus were not subsequently referred to hospitals) were not included because the diagnostic codes from these community-based general ophthalmology clinics are not reported to the DNPR.
Statistical analysis was performed from April 1, 2015, to October 31, 2016. We calculated descriptive statistics and presented the results as frequencies or mean (SD) values. The rates of AMD were calculated for all patients with MPNs as well as separately for each MPN subtype. The rates were compared with the rates of the matched controls and expressed as events per 1000 person-years at risk with 95% CIs. We computed the absolute risk of AMD for patients and controls during the first year, first 3 years, first 6 years, and first 10 years after the index date. The Kaplan-Meier method was used to estimate the cumulative incidence of AMD. We used the Cox proportional hazards regression model for the time-to-event analyses and the hazard ratio (HR), subsequently comparing the incidence of AMD observed among the patients with the incidence of AMD in the controls. Because smoking is a known risk factor for AMD32 and is suggested to be a risk factor for MPNs as well,33 we used smoking-associated diagnoses in the DNPR as a proxy measure for smoking. We lacked detailed patient information, such as their smoking status and specific treatments, because of the registry-based study design.
We initially performed stratified analyses using the periods of 1994-2001, 2002-2006, and 2007-2013 and adjusted the final analysis for risk time in these predefined periods. This stratification was done since new treatments were introduced for neovascular AMD at Danish hospitals in 2002 (verteporfin) and 2007 (anti-VEGF). Thus, persons with AMD were more likely to be referred to hospitals (for evaluation and potential treatment) rather than to general ophthalmology clinics in the most recent periods because these new treatments were not administered outside hospitals. Hence, more AMD events were recorded in the DNPR in the most recent period, regardless of exposure to MPNs.
In addition, we also estimated the risk of specifically coded neovascular AMD in an analysis confined to individuals with risk times after 2006 (without prior neovascular AMD). This analysis was performed since anti-VEGF treatment was administered nationwide at hospitals thereafter, and all individuals potentially eligible for treatment were evaluated at hospitals regardless of whether they had attended general ophthalmology clinics or hospitals before. Thus, no underreporting of neovascular events (requiring anti-VEGF treatment) occurred within that period.
The proportional hazard assumption was tested in all the models, and all statistical analyses were performed using the SAS statistical software package, version 9.4 (SAS Institute Inc), and R, version 3.2.2 (R Foundation for Statistical Computing). P < .05 was considered significant.
We identified 8323 persons who received a diagnosis of MPN and 83 200 individually matched controls. Of these, 365 patients and 5755 controls were excluded, including 222 patients with prior AMD and 143 patients with less than 30 days of follow-up. A higher percentage of patients with MPNs than controls had prior AMD (eTable 2 in the Supplement), and the odds ratio of prior AMD overall was 1.1 (95% CI, 1.0-1.3). As shown in Figure 1, controls corresponding to the excluded patients were also excluded from the study to maintain matching. In the final analysis, we included 7958 patients with MPNs (4279 women [53.8%] and 3679 men [46.2%]), with a mean (SD) age of 66.4 (14.3) years at diagnosis. The baseline characteristics are shown in Table 1, whereas the number of incident AMD events, according to the different periods and MPN subtypes, is shown in eTable 3 in the Supplement. During the study period, 234 patients with MPNs received a diagnosis of AMD, corresponding to a rate of 5.2 (95% CI, 4.6-5.9) per 1000 person-years at risk. The corresponding rate for the 77 445 controls was 4.3 (95% CI, 4.1-4.4). The rates across each MPN subtype and matched controls are shown in Table 2. The mean (SD) time to AMD was 5.6 (4.7) years for patients with MPN and 6.3 (4.9) years for the controls, with mean (SD) ages at AMD diagnosis of 77.9 (8.8) years for patients with MPNs and 79.1 (8.0) years for the controls. Specific results for MPN subtypes are shown in Table 2.
The absolute risk of AMD for patients with MPNs was 0.4% (95% CI, 0.2%-0.5%) during the first year, 0.9% (95% CI, 0.7%-1.2%) during the first 3 years, 1.7% (95% CI, 1.4%-2.0%) during the first 6 years, and 2.4% (95% CI, 2.1%-2.8%) during the first 10 years. The absolute risks of AMD for patients with MPNs and controls within each subtype are shown in Table 2. The cumulative risk revealed that the absolute risk of AMD increased over time for the patients with MPNs as well as for the general population (Figure 2). The risk was comparable for the patients with MPNs and their controls during the first few years, but the risk was higher for the patients with MPNs than for the controls hereafter.
In addition, our analyses revealed that the risk of AMD was significantly increased in the most recent periods and that smoking-related diagnoses were associated with an increased risk (Table 3). The crude HR of AMD for patients with MPNs vs controls was 1.3 (95% CI, 1.2-1.4), and the adjusted HR was 1.3 (95% CI, 1.1-1.5). The adjusted HRs for AMD were increased across the different MPN subtypes, with HRs of 1.2 (95% CI, 1.0-1.6) for essential thrombocythemia, 1.4 (95% CI, 1.2-1.7) for polycythemia vera, 1.7 (95% CI, 0.8-4.0) for myelofibrosis, and 1.5 (95% CI, 1.1-2.1) for unclassifiable MPNs.
There was a higher risk of neovascular AMD in the most recent period among patients with MPNs. Twenty patients with MPNs were recoded with specific-coded neovascular AMD (7 in essential thrombocythemia, 8 in polycythemia vera, 0 in myelofibrosis, and 5 in unclassifiable MPNs) after anti-VEGF treatment was introduced. The crude HR between patients with MPNs and controls was 1.3 (95% CI, 1.2-1.6), and the HR adjusted for smoking was 1.4 (95% CI, 1.2-1.6). Results for the MPN subtypes are shown in Table 3.
Using validated national registries, we included 7958 patients with essential thrombocythemia, polycythemia vera, myelofibrosis, or unclassifiable MPNs and 77 445 matched controls in a population-based observational study and found that patients with MPNs have an increased risk of AMD. To our knowledge, our study is the first to describe the risk of AMD in patients with MPNs, but we can only speculate on the causal pathways linking MPNs and AMD. The association may be explained in part by common inflammatory pathophysiological mechanisms because signs of chronic inflammation can be found in both patients with MPNs and patients with AMD.4,11 Thus, complement activation, increased oxidative stress responses, and elevated levels of inflammatory markers are common features in both diseases.17,19-26 Moreover, nuclear factor erythroid 2–related transcription signaling is impaired in hematopoietic stem cells as well as in retinal pigment epithelial cells.27,34 This signal plays a pivotal role in the regulation of the antioxidant responses and in the modulation of hematopoietic stem cell function.35
The etiological causes and the pathophysiology of MPNs are not fully known, but it is evident that the modulation of the Janus kinase–signal transducers and activators of transcription (JAK-STAT) pathway plays a central role. In the last decade, the understanding of the molecular pathogenesis has progressed markedly, and mutations in driver genes (JAK2 [GenBank NG_009904], CALR [GenBank NG_029662], and MPL [GenBank NG_007525]) that alter the signaling of the JAK-STAT pathway have been identified in patients with MPNs.36 In addition, immune modulations that promote the proliferation of the malignant stem cell clone in the bone marrow, which in turn increases systemic inflammation, are involved.11 Inflammation also plays a pivotal role in diseases that have previously been associated with MPNs, such as thrombosis,37 atherosclerosis,38 and autoimmune diseases.39
The pathophysiology is also not entirely known for AMD, but well-recognized risk factors include a family history of AMD, older age, and smoking.32,40,41 Immune mechanisms are also altered in patients with AMD,42 and systemic inflammatory changes seem to be involved in the development and progression of the disease.4,23,43,44 Like MPNs, AMD has previously been associated with systemic diseases that are characterized by chronic inflammation and/or immune deregulation, such as cardiovascular diseases,45,46 type 1 or type 2 diabetes,5 psoriasis,6 and AIDS.7 All of these features support our hypothesis that the increased risk of AMD in patients with MPNs may be explained by chronic systemic inflammation. It is intriguing to consider whether the malignant stem cell clone with concurrent chronic inflammation and immune deregulation might predispose some individuals to develop AMD.
Despite the nationwide setting and complete follow-up, our study has some limitations. First, not all AMD events were included, and our results did not depict the incidence of AMD in Denmark because the DNPR does not capture information on individuals who are followed up solely at general ophthalmology clinics. However, since new AMD treatments, which are administered exclusively at hospitals, were introduced for patients with neovascular AMD, and all patients were mandatorily referred for treatment evaluation at hospitals, accordingly, we expect that almost all neovascular AMD events in the most recent periods have been included in our study.
Second, because detailed patient-specific information was not available in our registry data, we could not validate the diagnoses or include variables such as smoking status and administrated MPN treatments (eg, phlebotomy, aspirin, or cytoreductive treatment). Distinguishing between different MPNs at diagnosis can be difficult in clinical practice47; hence, we cannot exclude the potential misclassification of patients across the MPN subtypes. This possibility implies some limitations in interpreting the MPN subtype results. However, the diagnostic codes for hematologic cancers have high positive predictive values in the DNPR.48 In our study, the influence of smoking was controlled using proxy measures, but residual confounding by smoking cannot be excluded. Moreover, because an association between aspirin use and the risk of AMD has been proposed,49,50 the use of aspirin could be a confounding factor in our study, as patients with MPNs are regularly treated with aspirin. However, a meta-analysis based on 10 studies, which included 171 729 individuals, found that aspirin use was not associated with AMD.51 Because both observational studies show opposing results regarding the association between aspirin use and AMD, and because experimental results are lacking, the evidence of an association is sparse. As suggested by others,49,52 we speculate whether the previously reported link between AMD and aspirin use is caused by confounding by indication—that is, that the underlying indication for aspirin treatment may increase the risk of AMD.
Third, vascular events can occur in patients with MPNs,53,54 and ophthalmic symptoms such as blurred vision and scintillating scotomas due to microcirculatory disturbances and transitory ischemia in the retinal vessels have been reported.55 Because patients with MPNs may present with blurred vision at diagnosis (which often resolves after the initiation of MPN-specific treatments such as phlebotomy, low-dose aspirin, and/or cytoreductive therapy),56-58 more of the patients than the controls may have had ophthalmologic evaluations. This possibility could have led to the coincidental finding of AMD in few patients with MPNs (eg, asymptomatic intermediate-stage AMD). However, we found only a few recorded AMD events in the month before or after the MPNs were diagnosed, and we minimized this bias by commencing follow-up after 30 days. We also found a higher prevalence of AMD at diagnosis, which indicates that the association between MPNs and AMD is not merely caused by referrals for ophthalmologic evaluation after the patients received a diagnosis of MPN. This finding strengthens the interpretation of our results, which suggests that MPNs are biologically associated with AMD, likely already at the time of diagnosis. Further substantiating our hypothesis, we found an increased risk of neovascular AMD in patients with MPNs after anti-VEGF treatment was introduced at hospitals nationwide—an era in which all neovascular events were recorded in the DNPR, irrespective of MPN exposure and whether persons had first attended general ophthalmology clinics or hospitals (some patients with late-stage AMD would likely have been followed up solely at general ophthalmology clinics before).
We find that patients with MPNs are at an increased risk for AMD when compared with age- and sex-matched controls from the general population. The association may possibly be explained by systemic inflammation. Our findings support that systemic changes may be involved in the development of AMD, but additional studies are needed to examine the underlying causal association.
Corresponding Author: Marie Bak, MD, Department of Haematology, Zealand University Hospital, University of Copenhagen, Sygehusvej 10, 4000 Roskilde, Denmark (firstname.lastname@example.org).
Accepted for Publication: May 8, 2017.
Published Online: June 22, 2017. doi:10.1001/jamaophthalmol.2017.2011
Author Contributions: Dr Flachs had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bak, Sørensen, Zwisler, Frederiksen, Hasselbalch.
Acquisition, analysis, or interpretation of data: Bak, Sørensen, Flachs, Juel, Frederiksen, Hasselbalch.
Drafting of the manuscript: Bak.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Bak, Flachs, Juel.
Obtained funding: Bak.
Administrative, technical, or material support: Bak.
Study supervision: Bak, Sørensen, Flachs, Zwisler, Frederiksen, Hasselbalch.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Frederiksen reported receiving research funding from Novartis. No other disclosures were reported.
Funding/Support: This study was supported by grants from the Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark; the Region Zealand Health Scientific Research Foundation; the Anders Hasselbalch Anti-Leukemia Foundation; and the A. P. Møller Foundation for the Advancement of Medical Science.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentations: Partial results were presented at the American Hematology Association Annual meeting; December 7, 2015; Orlando, Florida; and the European Hematology Association Annual Congress; June 2016; Copenhagen, Denmark.
Create a personal account or sign in to: