Causal effects of physical activity on the risk of overall ovarian cancer: A Mendelian randomization study

Abstract

Objective

Inconsistent results were reported on the association of physical activity with ovarian cancer. However, given the limitations of confounders and inverse causation, the validity of the association remained unclear. Therefore, we conducted a two-sample Mendelian randomization analysis, which can effectively avoid the aforementioned interference, to evaluate whether physical activity had a protective effect on ovarian cancer.

Methods

The exposure of interest was physical activity (both self-reported moderate-to-vigorous physical activity and accelerometer-measured physical activity). Summary statistics for physical activity traits were recruited from the UK Biobank (n = 91,084–377,234), whereas ovarian cancer summary genetic data were obtained from a genome-wide association study involving 25,509 cases and 40,941 healthy individuals. The inverse variance weighted approach was used as the primary Mendelian randomization method. Sensitivity analyses using Mendelian randomization-Egger regression, weighted median, and Mendelian randomization pleiotropy residual sum and outlier were also performed.

Results

The Mendelian randomization analyses indicated that there was no effect of moderate-to-vigorous physical activity (odds ratio, 1.11; 95% confidence interval: 0.66–1.85; P = 0.702), accelerometer-measured “average acceleration” (0.99 [0.91–1.08]; P = 0.848), and “overall activity” physical activity (0.97 [ 0.48–1.95]; P = 0.927) on the risk of overall ovarian cancer. However, “overall accelerations” physical activity (0.18 [0.05–0.64]; P = 0.008) were suggestively related to a lower risk of endometrioid ovarian cancer.

Conclusions

The Mendelian randomization analyses suggested that physical activity may not help to decrease the risk of overall ovarian cancer.

Keywords

Ovarian cancer Physical activity Mendelian randomization

Background

Ovarian cancer (OC) is the second most common and the most lethal gynecological malignancy.¹ It was estimated that approximately 313,959 newly diagnosed cases and 207,252 deaths to occur in 2020 worldwide.¹ As the symptoms of OC are often subtle and can be easily confused with symptoms of other less severe diseases, many patients were diagnosed at later stages with rather poor prognoses, and there was no effective screening method for OC currently.^2,3 Studies have shown that the five-year survival rate was as low as 30%–40%.^4,5 Therefore, the identification of modifiable protective factors, together with the development of more effective treatment, is of vital value to reduce disease burden.

Physical activity (PA), as one of the modifiable factors, has the potential to lower the risk of OC through its effects on ovulation, inflammatory markers, estrogen levels, and metabolic pathways.^6–8 However, this impact is inconclusive among studies. For instance, a pooled analysis of nine case-control studies, including 8,309 epithelial OC (EOC) patients and 12,612 controls, suggested that increasing PA was associated with a decreased risk of invasive OC.⁹ In contrast, the results from the European Prospective Investigation into Cancer and Nutrition (EPIC), involving 731 EOC cases and 274,740 healthy participants, did not support such an association.¹⁰ Despite inconsistent findings, the results from these observational studies suffered from many confounders such as obesity, which causes PA tougher, is positively related to OC.^11,12 In addition, these results may be disturbed by reverse causation, considering that patients with OC are prone to reduce PA.¹³

Mendelian randomization (MR) is an alternative method that utilizes single-nucleotide polymorphisms (SNPs) as a proxy for exposure to explore the causal relationship between exposures and outcomes.¹⁴ In comparison with traditional observational studies, MR can provide more robust evidence since genetic variants are randomly assigned at conception, avoiding confounders and reverse causation.¹⁵ Thus, in this study, we performed a two-sample MR analysis to evaluate the potential causal effects between PA and OC.

Methods

Genome-wide association studies summary statistics for PA

The summary statistics from two recently published genome-wide association studies (GWAS) on self-reported moderate-to-vigorous PA (MVPA) and accelerometry PA from the UK Biobank were leveraged.^16,17 In short, the UK Biobank is a large prospective cohort study of about a half-million participants with a wide range of measurements, including whole-genome markers.¹⁸ Self-reported MVPA was measured through a touchscreen questionnaire similar to the International PA Questionnaire.^16,19 Accelerometer-based PA was assessed by Axivity AX3 wrist-worn triaxial accelerometers.²⁰ Two measures were used for accelerometer-measured PA, of which accelerometer-measured “average acceleration” (overall acceleration average in milli-gravities) was examined from up to 7 days of accelerometer wear,¹⁶ whereas average vector magnitude for each 5-s epoch was used for the estimation of accelerometer-measured “overall activity” levels.²⁰

Instrumental variables selection for PA

The GWASs for MVPA and accelerometer-based PA identified nineteen and fifteen SNPs at a genome-wide significance level (P < 5 × 10⁻⁸) respectively.^16,17 Then we leveraged PLINK for SNPs independence (r² threshold = 0.001, window size = 10 mB). In the PhenoScanner database,²¹ four SNPs for accelerometer-measured were excluded owing to nominal association with OC (P < 1 × 10⁻⁵)(Supplemental Tables S1, S2). To investigate weak instrumental variable (IV) bias, we produced R² and F-statistic estimations. R² is the proportion of variance in exposure factors explained by IVs, and F-statistic < 10 suggests that the weak IV bias will be eliminated. The F-statistics of all SNPs included are greater than 10 (Supplemental Tables S3–S5). Ultimately, nineteen, eight, and, three SNPs were utilized as instruments for MVPA, "average acceleration" PA, and "overall activity" PA, respectively (Supplemental Tables S3–S5).

GWAS summary statistics for OC

To assess whether PA is related to OC, we identified genetic statistics from the EOC GWAS conducted by the Ovarian Cancer Association Consortium (OCAC).²² All 66,450 samples were of European descent, of which 25,509 were OCs.²² The numbers of histological subtypes of OC are presented in Supplemental Table S6.

Statistical analysis

We implemented the inverse-variance weighted (IVW) as our primary MR method.^23,24 Since the existence of horizontal pleiotropy could not be determined by IVW, we also performed other established MR methods with relatively robust estimation of horizontal pleiotropy, albeit at the cost of reduced statistical power. These methods included the weighted median method, which selects the median MR estimate as the causal estimate and gives a valid MR estimate even when up to 50% of IVs are invalid,²⁵ and MR-Egger regression, which estimates causal effects in the presence of pleiotropic effects.²⁶ In addition, MR pleiotropic residual sum and outliers (MR-PRESSO) was also applied to detect and correct for any outliers that might reflect pleiotropic bias in all reported results.²⁷ Cochran's Q statistic (P < 0.1) was performed to detect heterogeneity and Leave-one-out analysis was exploited to discover SNPs driving the outcomes (P < 0.05).^28–30 Results were presented as OR per 1 standard deviation increment in MVPA (8.14 milligravities) and accelerometer-measured PA (8 milligravities or 0.08 m/s²).^16,20 A Bonferroni-corrected P value was used, and values smaller than 0.007 (0.05/7) were identified as statistically significant. P values between 0.007 and 0.05 were considered as suggestive associations.³¹ All the analysis was performed using the TwoSampleMR packages and MR-PRESSO packages in R.

Results

MVPA and OC

For the 19 identified SNPs of MVPA, we conducted MR analyses to investigate the causal relationship of MVPA with OC and its histologic subtypes. The IVW method indicated that there was no association between MVPA and the risk of OC (odds ratio [OR], 1.11; 95% CI: 0.66–1.85; P = 0.702) or its histologic subtypes (Figures 1,2, Supplemental Tables S7, S8). We did not find any heterogeneity or horizontal pleiotropy among the SNPs using Cochran's Q test (Supplemental Table S9) and MR-Egger intercept test (Supplemental Table S10). In addition, other MR methods’ results were consistent with the IVW analysis (Figure 1). No single SNP made the MR estimates deviated (Supplemental Table S11).

Figure 1.

Mendelian randomization estimates for genetically determined physical activity on overall ovarian cancer. PA, physical activity; PRESSO, pleiotropy residual sum and outlier; OR, odds ratios; CI, confidence interval.

Figure 2.

Mendelian randomization estimates for genetically determined physical activity on endometrioid ovarian cancer. PA, physical activity; PRESSO, pleiotropy residual sum and outlier; OR, odds ratios; CI, confidence interval.

Accelerometer-measured PA and OC

Among the eight SNPs for “average acceleration” PA, we found that “average acceleration” PA was not related to the overall OC risk (IVW methods: 0.99[0.91–1.08], P = 0.848; Figure 1; Supplemental Table S12). The MR analysis revealed no significant effect of “overall activity” PA on the risk of overall OC (IVW method: 0.97 [0.48–1.95], P = 0.927; Figure 1; Supplemental Table S12). Heterogeneity was observed among SNPs of “average acceleration” PA (Supplemental Table S9), but the MR-Egger intercept test suggested no horizontal pleiotropy (Supplemental Table S10). Moreover, the leave-one-out analysis indicated that no single SNP dominated the MR estimates (Supplemental Tables S13, S14).

Consistently, we failed to discover any causal associations between “average acceleration” PA and any histological types of OC (Supplemental Tables S15, S16). However, we observed a suggestive association of “overall activity” PA with decreased risk of endometrioid OC (IVW: 0.18 [0.05–0.64]; P = 0.008) (Figure 2; Supplemental Tables S15, S17). No heterogeneity and pleiotropy were found (Supplemental Tables S9, S10). Furthermore, the leaving-one-out analyses revealed that no single SNP drove the results (Supplemental Table S14).

Discussion

In this study, a two-sample MR analysis was performed using summary statistics from two recently published GWASs and the OCAC to assess the causal relationship between PA and OC risk. There was no evidence that PA (both self-reported MVPA and accelerometer-measured PA) had a protective effect on the risk of overall OC. However, we found that accelerometer-measured PA may be a protective factor for endometrioid OC.

Previous observational studies focusing on PA and the risk of OC demonstrated inconsistent results.^10,32–34 A meta-analysis that included 26 studies suggested that a protective effect was more likely to be observed in case-control studies, whereas in cohort studies, this effect was prone to be insignificant or even the opposite.³⁵ There are several reasons for the prior studies’ contradictory findings. First, observational studies might control for different confounders leading to residual confounding. Furthermore, due to the retrospective character of case-control studies, the causality of PA on OC failed to validate. In addition, the varying definitions of PA might somewhat alter the results. In line with several cohort studies,^34,36–39 we failed to discover that PA can significantly reduce OC risk. Compared with traditional observational methods, MR analysis can effectively avoid the interference of confounding and reverse causation. In addition, the PA GWAS dataset and our analysis are based on a large sample cohort, the UK Biobank. Therefore, we expect that the reliability of our study was improved compared to previous observational studies based on a relatively smaller sample size.

For the histologic subtypes of OC, a suggestive prevention impact was observed between accelerometer-measured PA and endometrioid OC. A case-control study elucidated that a higher level of moderate recreational activity was significantly associated with decreased risk of endometrioid OC in accordance with our results.⁴⁰ Compared to other histologic types of OC, endometrioid OC had unique molecular profile. Some mutated genes, such as PTEN, CTNNB1, PIK3CA, KMT2D, KMT2B, PIK3R1, ARID1, and TP53, were notably discovered in endometrioid OC patients and microsatellite instability was more frequent.⁴¹ A meta-analysis suggested that PA resulted in a statistically significant decline of estradiol.⁴² Moreover, McTiernan et al. found that total PA was negatively related to the level of estrone, estradiol, and androstenedione.⁴³ Therefore, PA may prevent the development of endometrioid OC by reducing epithelial damage and mediating hormone levels. In our study, the preventive effect on endometrioid OC was identified by objectively measured PA rather than self-reported. As earlier studies revealed, there could exist disparities in two types of PA.^44–46 Self-reported PA was prone to be overstated compared with objective measurements.⁴⁶ In addition, accelerated measurements of “chip heritability” estimations were more capable of detecting significant relationships than self-reported PA.¹⁶

To conduct MR analysis, three core assumptions need to be satisfied.⁴⁷ In this study, the first assumption was assessed by linear regression of the PA on the genetic variants and calculating the F-statistic, and we found that all F-statistics were larger than 10 for included variants, suggesting that genetic variants used in our study were robustly associated with PA (P < 5 × 10⁻⁸, F-statistic > 10). The second hypothesis demands that the genetic instruments are independent of confounders of PA or OC (Supplemental Tables S1 and S2). To check this assumption, we leveraged PhenoScanner GWAS database to eliminate SNPs in relation to available confounders (P < 1 × 10⁻⁵). The third hypothesis is that genetic instruments affect OC exclusively through PA (Supplemental Tables S7–S9). Directly testing the third MR assumption can be challenging. Therefore, we performed the MR-Egger regression to potentially test the pleiotropy, and suggesting no pleiotropy, which indicated that the third assumption may not be violated.

The limitations of this study should be noted. First, selected SNP-based heritability for PA can only explain a small part of the complicated variance in PA. Furthermore, it is desired that exposure and outcome samples have similar gender distributions. Unfortunately, there are currently no publicly available gender-specific data for MVPA and accelerator-measured PA. More gender-specific data are favored, especially when it comes to sex-related diseases. Moreover, we only exploited three kinds of PA, whereas recent research has shown intricate correlations among various durations, frequencies, and intensities of PA on different types of disease.⁴⁸ For the analyses regarding “overall activity” PA and OC, only limited SNPs were included, and the results of PA on the prevention of endometrioid OC need to be interpreted in caution. To undertake further MR studies, more comprehensive PA, OC, and its subtype-specific GWAS data are required.

In conclusion, our study failed to provide sufficient evidence in favor of PA (either self-reported or accelerometer-measured) in the prevention of overall OC. Nevertheless, a suggestive association of "overall activity" PA with decreased risk of endometrioid OC was observed, suggesting the importance of objective measurements for PA in effect estimations. In addition, more gender-specific data are favored to obtain more reliable estimations, especially regarding sex-related diseases such as OC.

Supplemental Material

sj-docx-1-dhj-10.1177_20552076231162988 - Supplemental material for Causal effects of physical activity on the risk of overall ovarian cancer: A Mendelian randomization study

Supplemental material, sj-docx-1-dhj-10.1177_20552076231162988 for Causal effects of physical activity on the risk of overall ovarian cancer: A Mendelian randomization study by Jing Wang, Huanling Zhao, Jiahao Zhu and Minmin Jiang in Digital Health

Footnotes

Acknowledgments

The authors acknowledge the efforts of the consortia in providing high-quality GWAS resources for researchers. The instrumental variants of physical activities have been contributed by the UK Biobank，a large prospective cohort, and have been available at GWAS dataset https://klimentidis.lab.arizona.edu/content/data and https://doi.org/10.5287/bodleian:yJp6zZmdj. The ovarian cancer GWAS catalogue have been contributed by the Ovarian Cancer Association Consortium (OCAC) and have been available at https://doi.org/10.1038/s41467-020-14389-8.

Contributorship

Conceptualization, methodology, and writing-review and editing were handled by J.W., H.Z., J.Z, and M.J; formal analysis was conducted by J.W. and H.Z.; data curation was performed by J.W., H.Z., and J.Z.; writing-original draft preparation was done by J.W. and H.Z.; visualization was conducted by J.W., H.Z., and J.Z.; supervision was done by M.J.; project administration was done by M.J. All authors have read and agreed to the published version of the manuscript.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Guarantor

M.J was the guarantor.

ORCID iD

Minmin Jiang

Supplemental material

Supplemental material for this article is available online.

References

Sung

Ferlay

Siegel

, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021; 71: 209–249.

Kurman

Carcangiu

Herrington

, et al. WHO Classification of Tumours of Female Reproductive Organs. 2014.

Maringe

Walters

Butler

, et al. Stage at diagnosis and ovarian cancer survival: evidence from the international cancer benchmarking partnership. Gynecol Oncol 2012; 127: 75–82.

Siegel

Miller

Jemal

. Cancer statistics, 2018. CA Cancer J Clin 2018; 68: 7–30.

Allemani

Weir

Carreira

, et al. Global surveillance of cancer survival 1995-2009: analysis of individual data for 25,676,887 patients from 279 population-based registries in 67 countries (CONCORD-2). Lancet 2015; 385: 977–1010.

Olsen

Bain

Jordan

, et al. Recreational physical activity and epithelial ovarian cancer: a case-control study, systematic review, and meta-analysis. Cancer Epidemiol Biomarkers Prev 2007; 16: 2321–2330.

McTiernan

. Mechanisms linking physical activity with cancer. Nat Rev Cancer 2008; 8: 205–211.

Fair

Montgomery

. Energy balance, physical activity, and cancer risk. Methods Mol Biol 2009; 472: 57–88.

Cannioto

LaMonte

Risch

, et al. Chronic recreational physical inactivity and epithelial ovarian cancer risk: evidence from the ovarian cancer association consortium. Cancer Epidemiol Biomarkers Prev 2016; 25: 1114–1124.

10.

Lahmann

Friedenreich

Schulz

, et al. Physical activity and ovarian cancer risk: the European prospective investigation into cancer and nutrition. Cancer Epidemiol Biomarkers Prev 2009; 18: 351–354.

11.

Kohler

Garcia

Harris

, et al. Adherence to diet and physical activity cancer prevention guidelines and cancer outcomes: a systematic review. Cancer Epidemiol Biomarkers Prev 2016; 25: 1018–1028.

12.

Mendis

Davis

Norrving

. Organizational update: the world health organization global status report on noncommunicable diseases 2014; one more landmark step in the combat against stroke and vascular disease. Stroke 2015; 46: e121–e122.

13.

Lin

Edbrooke

Granger

, et al. The impact of gynaecological cancer treatment on physical activity levels: a systematic review of observational studies. Braz J Phys Ther 2019; 23: 79–92.

14.

Byrne

Yang

Wray

. Inference in psychiatry via 2-sample Mendelian randomization-from association to causal pathway? JAMA Psychiatry 2017; 74: 1191–1192.

15.

Smith

Lawlor

Harbord

, et al. Clustered environments and randomized genes: a fundamental distinction between conventional and genetic epidemiology. PLoS Med 2007; 4: e352.

16.

Klimentidis

Raichlen

Bea

, et al. Genome-wide association study of habitual physical activity in over 377,000 UK Biobank participants identifies multiple variants including CADM2 and APOE. Int J Obes (Lond) 2018; 42: 1161–1176.

17.

Doherty

Smith-Byrne

Ferreira

, et al. GWAS Identifies 14 loci for device-measured physical activity and sleep duration. Nat Commun 2018; 9: 5257.

18.

Sudlow

Gallacher

Allen

, et al. UK Biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med 2015; 12: e1001779.

19.

Craig

Marshall

Sjostrom

, et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 2003; 35: 1381–1395.

20.

Doherty

Jackson

Hammerla

, et al. Large scale population assessment of physical activity using wrist worn accelerometers: the UK Biobank study. PLoS One 2017; 12: e0169649.

21.

Kamat

Blackshaw

Young

, et al. Phenoscanner V2: an expanded tool for searching human genotype-phenotype associations. Bioinformatics 2019; 35: 4851–4853.

22.

Phelan

Kuchenbaecker

Tyrer

, et al. Identification of 12 new susceptibility loci for different histotypes of epithelial ovarian cancer. Nat Genet 2017; 49: 680–691.

23.

Chen

, et al. Waist circumference increases risk of coronary heart disease: evidence from a Mendelian randomization study. Mol Genet Genomic Med 2020; 8: e1186.

24.

Burgess

Butterworth

Thompson

. Mendelian Randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol 2013; 37: 658–665.

25.

Bowden

Davey Smith

Haycock

, et al. Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol 2016; 40: 304–314.

26.

Bowden

Davey Smith

Burgess

. Mendelian Randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 2015; 44: 512–525.

27.

Verbanck

Chen

Neale

, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50: 693–698.

28.

Burgess

Thompson

. Interpreting findings from Mendelian randomization using the MR-Egger method. Eur J Epidemiol 2017; 32: 377–389.

29.

Greco

Minelli

Sheehan

, et al. Detecting pleiotropy in Mendelian randomisation studies with summary data and a continuous outcome. Stat Med 2015; 34: 2926–2940.

30.

Grover

. Mendelian Randomization: Methods for Using Genetic Variants in Causal Estimation. S. Burgess and S. G. Thompson (2015). London, UK: Chapman & Hall/CRC Press. 224 pages, ISBN: 9781466573178. Biometrical Journal 2017; 59.

31.

Yuan

Mason

Carter

, et al. Homocysteine, B vitamins, and cardiovascular disease: a Mendelian randomization study. BMC Med 2021; 19: 97.

32.

Rossing

Cushing-Haugen

Wicklund

, et al. Recreational physical activity and risk of epithelial ovarian cancer. Cancer Causes & Control : CCC 2010; 21: 485–491.

33.

Schnohr

Grønbaek

Petersen

, et al. Physical activity in leisure-time and risk of cancer: 14-year follow-up of 28,000 Danish men and women. Scand J Public Health 2005; 33: 244–249.

34.

Weiderpass

Margolis

Sandin

, et al. Prospective study of physical activity in different periods of life and the risk of ovarian cancer. Int J Cancer 2006; 118: 3153–3160.

35.

Cannioto

Moysich

. Epithelial ovarian cancer and recreational physical activity: a review of the epidemiological literature and implications for exercise prescription. Gynecol Oncol 2015; 137: 559–573.

36.

Patel

Rodriguez

Pavluck

, et al. Recreational physical activity and sedentary behavior in relation to ovarian cancer risk in a large cohort of US women. Am J Epidemiol 2006; 163: 709–716.

37.

Hannan

Leitzmann

Lacey

Jr , et al. Physical activity and risk of ovarian cancer: a prospective cohort study in the United States. Cancer Epidemiol Biomarkers Prev 2004; 13: 765–770.

38.

Bertone

Willett

Rosner

, et al. Prospective study of recreational physical activity and ovarian cancer. J Natl Cancer Inst 2001; 93: 942–948.

39.

Pukkala

Poskiparta

Apter

, et al. Life-long physical activity and cancer risk among Finnish female teachers. Eur J Cancer Prev 1993; 2: 369–376.

40.

Pan

Ugnat

Mao

. Physical activity and the risk of ovarian cancer: a case-control study in Canada. Int J Cancer 2005; 117: 300–307.

41.

Pierson

Peters

Chang

, et al. An integrated molecular profile of endometrioid ovarian cancer. Gynecol Oncol 2020; 157: 55–61.

42.

Ennour-Idrissi

Maunsell

Diorio

. Effect of physical activity on sex hormones in women: a systematic review and meta-analysis of randomized controlled trials. Breast Cancer Res 2015; 17: 39.

43.

McTiernan

Chen

, et al. Relation of BMI and physical activity to sex hormones in postmenopausal women. Obesity (Silver Spring) 2006; 14: 1662–1677.

44.

Prince

Adamo

Hamel

, et al. A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review. Int J Behav Nutr Phys Act 2008; 5: 56.

45.

Vancampfort

Firth

Schuch

, et al. Sedentary behavior and physical activity levels in people with schizophrenia, bipolar disorder and major depressive disorder: a global systematic review and meta-analysis. World Psychiatry 2017; 16: 308–315.

46.

Dyrstad

Hansen

Holme

, et al. Comparison of self-reported versus accelerometer-measured physical activity. Med Sci Sports Exerc 2014; 46: 99–106.

47.

Rasooly

Patel

. Conducting a reproducible Mendelian randomization analysis using the R analytic statistical environment. Curr Protoc Hum Genet 2019; 101: 82.

48.

Chekroud

Gueorguieva

Zheutlin

, et al. Association between physical exercise and mental health in 1.2 million individuals in the USA between 2011 and 2015: a cross-sectional study. Lancet Psychiatry 2018; 5: 739–746.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB