Factor V Leiden, estrogen, and multimorbidity association with venous thromboembolism in a British-South Asian cohort

Summary Multimorbidity, estrogen use, and Factor V Leiden (FVL) are known independent risk factors for venous thromboembolism (VTE). This cross-sectional analysis of women in the Genes & Health British-South Asian cohort (N 20,048) linked the F5 SNP rs6025 with estrogen prescribing data and VTE events. Multivariable logistic regression was used to test the association between estrogen use, FVL, common medical co-morbidities, and VTE. Estrogens were prescribed to 30% of women. 3% of participants were FVL carriers. 439 participants had a VTE event (2.2%), and VTE prevalence increased with obesity, hypertension, dyslipidemia, chronic kidney disease, estrogen use, and in the presence of FVL. One medical condition above was independently associated with VTE with an OR 1.6 (CI 1.2–2.0, p 0.001); two medical conditions OR 2.7 (CI 2.0–3.7, p < 0.001); three OR 5.3 (CI 3.8–7.4, p < 0.001); four OR 8.1 (CI 4.9–13.0, p < 0.001). Multimorbidity and FVL compound risk of VTE with estrogen use.


INTRODUCTION
3][4] Mortality from pulmonary embolism is significant and noted recently to be up-trending among younger patients (25-64 years old) in the USA. 5,6strogen containing oral combined contraception (OCP) and hormone replacement therapy (HRT) are commonly used by pre-and postmenopausal women.Data from the USA suggests that more than 80% of sexually active women had taken oral contraceptives, almost all of which were combination therapies containing estrogen. 7,8OCPs and Estrogen containing HRT are known to increase the relative risk of venous thromboembolism (VTE) significantly, though the absolute risk of VTE remains low. 9,10actor V Leiden (FVL) is caused by a single nucleotide polymorphism in the F5 gene (1691G>A substitution) and leads to a prothrombotic state, which has a synergistic increase in VTE risk with exogenous estrogen use. 11,12The mechanism is a resistance to activated protein C, which is an endogenous anticoagulant. 11Prevalence of FVL is known to vary across trans-ancestral groups, with highest prevalence in European ancestry and a lower prevalence in Asian populations. 13,14olymorbidity is increasing, and it remains unclear how intersection of multiple common medical conditions, exogenous estrogen use, and FVL may alter risk of VTE. 15 To address this gap in knowledge we analyzed the Genes & Health (G&H) cohort of Bangladeshi and Pakistani ancestry participants in the United Kingdom (UK).Although Asian populations are known to have lower prevalence of the FVL allele, as compared with European ancestry populations, the G&H cohort suffers from high rates of cardio-metabolic morbidity and a large percentage of women are likely to be exposed to exogenous estrogen across their lifetime.It is also a population that is grossly under-represented in clinical and preclinical research cohorts.
Pharmacogenomic panels are being considered for routine use in national clinical care in the United Kingdom's National Health Service, therefore revisiting utility of F5 pharmacogenomic testing to inform choice of contraception and HRT is timely. 16Pharmacogenomic panel testing shifts the issue of FVL testing from a population screening question and reframes it as a medicine optimization tool.Furthermore, prior health economic models used to estimate cost of genetic testing are obsolete in this context.
The aim of this study is to clarify how intersection of multiple common medical conditions, exogenous estrogen use, and FVL contribute to cumulative risk of VTE in a British South-Asian ancestry cohort.

VTE events
The relative risk of VTE in women carrying a FVL mutation who had been prescribed estrogen was more than double women who did not have a FVL mutation (4.6% vs. 2.1%, significant on fisher's exact testing p 0.047, OR 2.2, 95% CI 0.9-4.9).The majority of the 439 VTE events were phlebitis and thrombophlebitis (76%).Of those women prescribed estrogens, 21% of the participants with VTE (27/129) had a pulmonary embolism (0.5% of all women prescribed estrogen) (Table 2).

Multimorbidity
Prevalence of common medical comorbidities in the cohort prescribed estrogens are shown in Table 2.Those who had been prescribed estrogens were young at enrollment (mean age 37), with 21% obesity, 11% diabetes mellitus, 11% primary hypertension (HTN), 11% dyslipidemia, and 2% chronic kidney disease (CKD).Those prescribed estrogens were significantly younger and less likely to have a diagnosis of obesity, diabetes mellitus, primary hypertension, dyslipidemia, or chronic kidney disease as compared with the cohort who had not been prescribed estrogens.However, there was no significant difference in FVL prevalence between the two groups.
When VTE events were stratified by poly-morbidity status, prevalence of VTE increases with number of conditions (Table 5).Our results show that the absolute risk of VTE in this cohort is not trivial in those women with two or more co-existent medical conditions in the absence of Estrogen use, $4% with two conditions, rising to $14% with four co-morbid conditions.
This risk is amplified with the same trends in those proscribed estrogens, though a smaller percentage of participants with multiple comorbidities were prescribed estrogen compared with the baseline population (Table 5).In the sub-cohort prescribed estrogens who carry an FVL mutation there was an increase in VTE prevalence affecting those with more than one medical co-morbidity disproportionately (Table 5; Figure 1).19% of those carrying an FVL mutation with two medical co-morbidities had a VTE event (compared with $5% VTE prevalence in those prescribed estrogens and having 2 medical co-morbidities overall) (Table 5).Likewise, in a dose-dependent fashion, those with three medical co-morbidities who carried an FVL mutation had a 29% prevalence of VTE (compared with $9% in those with 3 medical co-morbidities prescribed Estrogen overall) (Table 5; Figure 1).

DISCUSSION
Our study shows an independent, statistically significant, and clinically meaningful increase in VTE prevalence in women who have FVL, had been prescribed estrogen, or had a diagnosis of obesity, HTN, CKD, or dyslipidemia.We demonstrated a cumulative significant association with VTE where several of these medical co-morbidities was present in combination, ranging from an OR 1.6 for one condition (CI 1.2-20, p 0.001) to OR 8.1 (CI 4.9-13.0,p < 0.001) for a participant with all four identified medical co-morbidities (not an uncommon patient to encounter in clinical practice) (Figure 2).This is the first such study to look at cumulative risk of common medical conditions, estrogen use and FVL on VTE prevalence in a South Asian ancestry western population.While independently these factors have all been associated with VTE to various degrees, prior studies have not aggregated commonly co-occurring medical conditions.Furthermore, South Asian ancestry populations in western countries are known to suffer from high rates of cardiometabolic morbidity. 17ur results show that the prevalence of VTE in this cohort is not trivial in those women with two or more co-existent medical conditions; rising to almost 1 in every 6 women with four comorbid conditions in the absence of Estrogen use.In the sub-cohort who had been prescribed estrogen the prevalence of VTE was nearly 1 in 5 for those women with all four medical conditions.In the presence of FVL, the prevalence of VTE with three medical conditions was nearly 1 in every 3 women.These absolute risks argue against prior dogma which resulted in a decision not to offer testing for FVL prior to estrogen prescription. 9early 1 in every 3 women in this study cohort had been prescribed estrogens.The high prevalence of exposure to estrogens emphasizes the importance of elucidating multifactorial VTE risk.
VTE risk is known to be multifactorial, with inherited, acquired, and environmental risk factors.However, the contribution of multimorbidity with chronic and commonly overlapping cardiovascular and metabolic conditions to VTE has not been well studied.It is important to elucidate the cumulative impact of multimorbidity with exogenous estrogen use and FVL to optimize informed choice of estrogen containing medication use.Future studies should explore the impact of overlap in multimorbidity, FVL, and Estrogen use in other geographic, socioeconomic and ancestral populations.Further work should be done to understand the various aspects of multi morbidity that may be contributing to VTE risk such as lifestyle habits and environmental exposures associated with the studied medical conditions.
Prior concerns were raised about women being denied contraception due to detection of FVL. 9 However, there are many safe and effective non-estrogen containing choices for contraception and an estimation of non-trivial VTE risk does not need to be a contraindication to use.In fact, there is an increasing emphasis on wholistic decisions making rather than treating all thrombophilia as contraindications to OCP use. 18F5 inclusion in a pharmacogenomic panel medicines optimization approach could therefore enable more personalized risk assessment and enable patients to make more informed decisions (Figure 3).
The population morbidity landscape has changed in the past several decades, as have contraception options and doctor patient decision making models.People are living longer but with more of life lived with co-morbidities. 19Projections suggest this will continue, with individuals experiencing four or more diseases estimated to reach 17% by 2035. 20Women are having children later in life and are more likely to have medical comorbidities during reproductive years than in the past. 21,22Contraception options have expanded, and there is now more emphasis on shared decision making and informed consent.The healthcare provider landscape is shifting toward pre-emptive pharmacogenomic testing for commonly used non-specialist prescribed medications: The PREPARE trial demonstrated reduction of ADRs by 30% using a panel approach in European centers.4][25] Therefore, a future when pre-emptive pharmacogenomic panels are readily available in routine care may be near at hand.These factors combined mean that a large number of women who are likely to consider taking exogenous estrogens in their lifetime may have an elevated baseline risk of VTE due to co-morbidities and other multifactorial additional risk factors, and that pharmacogenomic panel information may be available to help inform personalized discussion of VTE risk.Including F5 in such a panel would continue the shifting ideology of medicines optimization and shared decision making based on informed consent.

Limitations of the study
Due to the overall low prevalence of VTE and of FVL, the number of women who had an FVL allele, had been prescribed estrogen, and had a VTE event was small (7 women).Therefore, it would be ideal to replicate these data in a larger cohort.Furthermore, due to limitations of the data available, this is a cross-sectional study.This approach and lack of longitudinal data are likely to decrease our signal and mean that we are underestimating the effect of estrogen on VTE and accounts for the lower OR associated with VTE from estrogen use in our study versus prior studies.However, this biases our model against signal detection, it does not compromise the validity of the significant associations we have presented.Furthermore, though the medical conditions considered could plausibly lead to increased VTE risk, there is not a plausible pathway for VTE to lead to the occurrence of these medical comorbidities.We also did not analyze drug-drug interactions.

Clinical implications
As multimorbidity increases, it is important to examine cumulative risk of VTE from multiple common medical conditions, aging, and genetic risk prior to prescribing estrogen.FVL disproportionately increases VTE risk for those with multiple common medical co-morbidities taking estrogen contained in oral contraceptives due to additive risk.If these results are validated in other cohorts, it would suggest that not only obesity and HTN, but also dyslipidemia and chronic kidney disease should be considered and possibly even screened for before initiating estrogen therapy.Our cohort data suggests that clinicians are already less likely to prescribe estrogens to multimorbid patients, but that they are not less likely to prescribe estrogens to those with FVL (as it is not clinical practice to test for FVL in the absence of an unexplained thrombotic event or family history).This suggests the practice of asking about family history of VTE prior to prescribing estrogen is not significantly decreasing the percentage of patients with FVL being prescribed estrogens.
Figure 1.Prevalence of VTE Prevalence of VTE in women increases with increasing number of co-existent medical comorbidities identified on the x axis (obesity, hypertension, chronic kidney disease, dyslipidemia).There is an increase in VTE in those women prescribed estrogens that follows the same trend, increasing with number of medical comorbidities.In those prescribed estrogen and carrying a Factor V Leiden mutation there is a steep increase in VTE risk in those women with more than one medical co-morbidity.

Figure 2 .
Figure2.Study results overview 20, 048 UK women of Pakistani and Bangladeshi ancestry were analysed.Multivariable logistic regression was used to find the associations between estrogen use, FVL, common medical co-morbidities, and VTE.Estrogens were prescribed to 30% of women.3% of participants were FVL carriers.VTE prevalence increased independently with obesity, hypertension, dyslipidemia, chronic kidney disease, estrogen use, and in the presence of FVL.Multiple of the common medical co-morbidities above (obesity, hypertension, dyslipidemia, chronic kidney disease) resulted in escalating risk of VTE independent of FVL or oestrogen use.Multimorbidity and FVL compound risk of VTE with estrogen use.OR, odds ratio, CI, confidence interval.

Table 2 .
Female cohort characteristics and VTE events p values from fisher's exact test for discrete variables and t-test for continuous variables.*p value < 0.05, **p value < 0.001.

Table 4 .
Multivariable logistic regression testing for association between VTE and multiple of the 4 common medical conditions found to be significant above (obesity, HTN, CKD, dyslipidemia)

Table 5 .
Multimorbidity impact The above table outlines VTE prevalence in participants with increasing numbers of 4 common medical co-morbidities associated with VTE in this cohort: Obesity, hypertension, dyslipidemia, chronic kidney disease.Statistically significant difference with p value < 0.05 by Fisher's exact test in comparison with the column to the left noted by *. ** denotes p value < 0.001.