Epidemiologic Analysis Along the Mevalonate Pathway Reveals Improved Cancer Survival in Patients Who Receive Statins Alone and in Combination With Bisphosphonates

Abstract

Purpose

Cohort studies report associations between statin use and improved survival in patients with cancer. We used pharmacoepidemiologic methods to evaluate the survival of patients with cancer who received statins alone or in ostensibly synergistic drug combinations.

Materials and Methods

Patients with cancer who were diagnosed from 2010 to 2013 were identified in a large health care claims database. The rate of all-cause death up to 1 year after diagnosis was compared by Cox proportional hazard regression. Sensitivity analyses included age stratification, statin type and intensity, and comparison with or without bisphosphonates and dipyridamole.

Results

Among 312,907 identified patients with cancer, treatment groups included statin users (n = 65,440), nonstatin users who received medications that block cholesterol absorption (n = 9,289), and nonusers (n = 226,007). Statin use before diagnosis was associated with improved overall survival compared with no treatment (hazard ratio [HR], 0.85; 95% CI, 0.80 to 0.91) and specifically in patients with leukemia, lung, or renal cancers. Nonstatin users had increased overall survival compared with no treatment (HR, 0.73; 95% CI, 0.62 to 0.85); when stratified, this difference held true only for pancreatic cancer and leukemia. No differences were observed between statin and nonstatin groups. Bisphosphonate use alone had no effect (n = 4,528), but patients who used both statins and bisphosphonates (n = 4,090) had increased survival compared with no treatment (HR, 0.60; 95% CI, 0.45 to 0.81). The effect of the combination of dipyridamole and statin use (n = 651) was not significant compared with no treatment.

Conclusion

This study suggests that the combination of statins with drugs that affect isoprenylation, such as bisphosphonates, improves survival in patients with cancer. Consideration of pathway-specific pharmacology allows for hypotheses testing with the pharmacoepidemiologic approach. Prospective evaluation of these findings warrants clinical investigation and preclinical mechanistic studies.

INTRODUCTION

The role of 3-hydroxy-3-methylglutaryl coenzyme-A (HMG-CoA) reductase inhibitors (ie, statins) in chemoprevention and cancer treatment has been deliberated in the literature,^1-5 as has the mechanism by which statins may exert their effect and improve overall survival in patients with cancer.^6-9 Recent cohort studies have shown that current statin use is associated with significantly lower risk of death as a result of cancer,³ whereas other studies have shown that statin use after diagnosis can reduce cancer-specific mortality in patients with breast,¹⁰ colon,^11,12 or lung cancer.¹³ However, some skepticism remains, because mitigation of all confounding factors is not possible with cohort or pharmacoepidemiologic studies.^14,15 Prospective evaluation of statins as monotherapy in cancer has been attempted, but a recent review of clinical trials of statin monotherapy in cancer revealed little effect.¹⁶ In contrast, prospective trials that use a combination of statin with chemotherapy have shown improved survival in patients with cancer.^17-21 Given the relative safety profile of statins, rational combination therapies may provide patients with cancer clinical benefit at the expense of minimal toxicity risk.

Mechanistically, statins inhibit HMG-CoA reductase, the rate-limiting enzyme of the mevalonate pathway that ultimately produces cholesterol along with isoprenoids and intermediate pathway metabolites, such as farnesyl pyrophosphate (FPP) and geranylgeranyl pyrophosphate (GGPP).^22,23 The physiologic importance of isoprenoids in normal and cancer cells is critical for facilitation of membrane anchoring of numerous signaling molecules, including G proteins, such as Ras and Rho.²⁴ Furthermore, constitutive activation of signaling pathways in cancer is dependent on cholesterol availability for formation of lipid rafts.^25,26 Recognition of the biologic importance of FPP and GGPP led to drug development efforts of isoprenylation enzyme inhibitors^27,28; these efforts ultimately stalled in early-phase studies because of toxicity and lack of efficacy. Pharmacologically, bisphosphonates also act as isoprenylation inhibitors^29,30 and are widely used to inhibit bone resorption and treat osteoporosis. Bisphosphonates also have been used extensively to treat patients with cancer who have bone metastases.^31-34 Their effect on reduction of bone metastasis is significant and is consistent with the observations that they cause apoptosis in tumor cells via inhibition of the mevalonate pathway as well as tumor cell invasion in vitro.^35,36 Reduced cholesterol levels trigger a negative feedback loop, which ultimately leads to the upregulation of mevalonate pathway genes via the transcriptional activity of sterol regulatory element binding protein transcription factors.^37,38 On the basis of this understanding, preclinical studies have demonstrated synergy when statins are combined with bisphosphonates^6,35,39-43 and, most recently, with dipyridamole, which has been shown to inhibit sterol regulatory element binding protein 2.^44,45 Thus, the use of combination therapies that can amplify the effect of statins or abrogate the development of statin-related resistance may lead to synergistic clinical combinations. Here, we used a large data set of health claims to test the hypothesis that statins alone or in combination with potentially synergistic therapies prolong the survival of patients with cancer. The combination of epidemiologic evidence and preclinical data may provide a strong rationale for future prospective clinical studies.

MATERIALS AND METHODS

We used the Truven Health MarketScan Commercial Claims and Medicare Supplemental Databases (Ann Arbor, MI). The MarketScan database includes approximately 40 million individuals from more than 160 large employers and health plans across the United States and includes health care claims with diagnosis and procedure codes for medical encounters and all prescription medication fills. Data were de-identified in compliance with the Health Insurance Portability and Accountability Act regulations, and the University of Kentucky institutional review board approved the use of the database for this study.

Patient Selection

Adults ≥18 years of age diagnosed with cancer between January 1, 2010, and November 31, 2013, were identified by using International Classification of Diseases (9th revision [ICD-9]) codes in the primary or secondary positions. Patients with prostate and breast cancer were excluded, because the use of hormonal therapy affected the risk for thromboembolism. Eligible patients were diagnosed with one of the following types of cancer: stomach (ICD-9 codes: 151.xx), pancreatic (157.xx), brain (191.xx), lung (162.2 to 162.9), renal (189.0, 189.1), lymphoma (200.xx to 202.xx), leukemia (204.xx to 208.xx), myeloma (203.0x), colorectal (153.xx, 154.xx), or gynecologic (179.xx, 180.xx, 182.xx, 183.xx). At least two inpatient or outpatient diagnoses within 14 days were required, and the date of the first qualifying diagnosis of cancer was defined as the index date. Patients were required to have at least 12 months of preindex and a minimum of 1 month postindex continuous enrollment in the database.

Exposure groups were defined as statin users with no history of nonstatin cholesterol-lowering medication use; nonstatin cholesterol-lowering medication users with no history of statin use (nonstatin users, the active control group); and those with no history of statin or nonstatin medication use (nonusers, the control group). Medication use was based on at least 90 cumulative days of medication supplied in the 6 months before diagnosis. Specific statins include lovastatin, pravastatin, simvastatin, rosuvastatin, atorvastatin, fluvastatin, and pitavastatin. Nonstatins include fibric acid derivatives, bile acid sequestrants, and nicotinic acid.

Measures

Patient demographic characteristics included age, sex, geographic region, and urban residence. Clinical characteristics measured during the 12-month preindex period included the Charlson Comorbidity Index (CCI) and Elixhauser Comorbidities Index.^46,47 These include 17 and 31 categories of comorbid conditions, respectively, and are widely used for risk adjustment with health outcomes data. Additional medications accounted for in the preindex period included anticoagulants, antihypertensives, antiplatelet, antiarrhythmics, and digoxin. Presence of metastatic disease was assessed on the index date.

Statistical Methods

Pairwise analyses were done between statins versus nonstatins, statins versus nonusers, and nonstatins versus nonusers. Propensity-score matching was conducted with baseline comorbidities, medications, and demographic information to achieve balance between treatment groups. Propensity scoring mimics the random assignment process of a clinical trial so that each matched pair has the same baseline probability to receive either treatment.^48-50 Matched pairs should be similar in all baseline characteristics. Patients with the same cancer type were matched by using a greedy, nearest neighbor algorithm with a caliper set at 0.2 times the standard deviation of the propensity scores in the sample, which allowed for up to five matches for each treated person.⁵¹ Standardized differences were calculated and are shown in the Data Supplement. A standardized difference of < 0.10 generally was considered nonsignificant.⁵² To address any residual confounding after propensity-score matching, covariates also were incorporated in the final regression models.⁴⁹ The final model included the following adjustment covariates: age, CCI score, region, anticoagulants, antihypertensives, antiplatelets, antiarrhythmics, digoxin, Elixhauser Comorbidities Index, and preindex history of coronary heart disease, deep vein thrombosis, pulmonary embolism, atrial thrombosis, and myocardial infarction. Two sample t test and χ² tests were conducted to assess significant differences between treatment groups before and after matching.

The study cohort was observed until patients died or were lost to follow-up because of loss of enrollment in the data set, or until the end of the study data. Cox proportional hazard regression models, which accounted for correlation within matched pairs, were used to assess risk of death within 1 year of diagnosis among all cancers and then were stratified among each cancer. Follow-up was terminated for those who survived beyond 1 year, and data were censored. Hazard ratios (HRs) and 95% CIs were reported. A P value of < .05 was considered statistically significant. All matching and statistical analyses were conducted in SAS (SAS Institute, Cary, NC).

Subgroup Analyses

A sensitivity analysis to assess the influence of age as an effect modifier was completed. Comparisons were made within age groups of younger than 65 years, 65 to 75 years, and older than 75 years, because these groups may have different treatment patterns, responses, and baseline survival before cancer diagnosis. The effect of dose intensity was evaluated by limiting the statin cohort to moderate- or high-dosage statins and by comparing results again with nonstatins and nonusers.⁵³ Patients who remained on statin therapy after cancer diagnosis were assessed by observing the proportion of days in which medication was available to the patient in the postindex period until the end of follow-up. Statins were compared individually and by type to ascertain differences among outcomes. Natural statins were lovastatin, pravastatin, and simvastatin; synthetic statins were rosuvastatin, atorvastatin, fluvastatin, and pitavastatin.⁵⁴ Patients who used bisphosphonates or dipyridamole, alone or in combination with statins, were compared with nonusers to assess effectiveness or synergy in terms of survival. Bisphosphonates assessed in the analysis were alendronate sodium, etidronate disodium, ibandronate sodium, pamidronate disodium, risedronate sodium, and zoledronic acid. Bisphosphonates and dipyridamole were subject to the same requirement of having at least 90 cumulative days supplied in the 6 months before cancer diagnosis. For all subgroup analyses, all cohorts were rematched via propensity-score methods.

RESULTS

Table 1 lists the baseline demographic and clinical characteristics of the study population by treatment group. Because of the enrollment criteria, there were no missing values on covariates used for propensity-score matching or survival analyses. Eligibility criteria were met by 312,907 patients with cancer. Three treatment groups were established as outlined in the CONSORT diagram (Fig 1): statin users (n = 65,440), nonstatin users who received nonstatin cholesterol-lowering medications (n = 9,289), and nonusers (n = 226,007). A total of 8,198 patients died within 1 year of diagnosis; 1,702 patients were from the statin-users cohort, 216 were from the nonstatin-users cohort, and 6,280 were from the no-treatment cohort. The cohort contributed an average follow-up time of 359 days (standard deviation [SD], 39.3 days), and there were no differences between cancer types or treatment groups. The mean (SD) ages of patients in the statin, nonstatin, and nonuser cohorts were 74.2 years (7.8 years), 71.8 years (9.3 years), and 60.8 years (14.1 years), respectively. In all treatment groups, lung cancers, lymphomas, and colorectal cancers accounted for the top three diagnosed cancers, whereas stomach, pancreatic, and brain cancers were the least diagnosed. The numbers of matched pairs after propensity-score matching were 39,989, 101,401 and 27,319 for the comparisons of statin users versus nonstatin users, statin users versus no treatment, and nonstatin users versus no treatment, respectively. Although baseline differences existed among treatment groups, matching provided samples that had minimal differences (Data Supplement).

Table 1.

Baseline Demographic and Clinical Characteristics of All Cohorts Included in the Analysis

Fig 1.

Patient flow diagram.

Figure 2 displays the HR and associated 95% CI for the effect of treatment group on survival within the propensity-matched sample. Overall, there were no differences in survival between statin users and nonstatin users. Among all cancers, statin use before diagnosis was associated with improved overall survival compared with no treatment (HR, 0.85; 95% CI, 0.80 to 0.91). When stratified by cancer type, this observation held true for lung cancer (HR, 0.88; 95% CI, 0.78 to 0.98), renal cancer (HR, 0.63; 95% CI, 0.44 to 0.90), and leukemia (HR, 0.73; 95% CI, 0.58 to 0.92). Nonstatin use provided a similar reduction in overall survival compared with no treatment (HR, 0.73; 95% CI, 0.62 to 0.85); however, when stratified, this difference held true only for pancreatic cancer (HR, 0.53; 95% CI, 0.29 to 0.98) and leukemia (HR, 0.53; 95% CI, 0.30 to 0.94).

Fig 2.

The 1-year survival analysis. Hazard ratios (HRs) and associated 95% CIs for the effect of treatment group on survival within the propensity-matched sample. Results included the entire treated group (ie, all selected cancers) and groups stratified by cancer type. Propensity-score methods were used to reduce the effects of confounding to mimic randomized controlled trials. The full multivariable Cox model included the following adjustment covariates: age; Charlson Comorbidity Index score; region; uses of anticoagulants, antihypertensives, antiplatelets, antiarrhythmics, or digoxin; Elixhauser Comorbidities Index; and pre-Index history of coronary heart disease, deep vein thrombosis, pulmonary embolism, atrial thrombosis, or myocardial infarction. Matching was conducted for each treatment-group analysis.

Dose-intensity analysis determined that the effect observed was not dosage dependent (Fig 3). After removal of low-dosage statins and comparison to nonstatins and nonusers, overall survival differences were of the same magnitude and direction observed in the overall analysis. The sensitivity analysis to assess age (Table 2) also showed no difference in effect within age groups when statins were compared with nonstatins or nonstatins compared with nonusers. However, patients younger than age 65 years in the comparison of statin users versus nonusers had improved survival, whereas those older than age 65 years did not.

Fig 3.

The effect of dose intensity on 1-year survival. Statin exposure groups were limited to medium (med) and high dosages. The distributions of low-, moderate-, and high-dosage statins before matching were 7,918, 46,152, and 11,307, respectively. Hazard ratios (HRs) and associated 95% CIs for the effect of treatment group are presented. The full multivariable Cox model included the following adjustment covariates: age; Charlson Comorbidity Index score; region; uses of anticoagulants, antihypertensives, antiplatelets, antiarrhythmics, or digoxin; Elixhauser Comorbidities Index; and pre-Index history of coronary heart disease, deep vein thrombosis, pulmonary embolism, atrial thrombosis, or myocardial infarction. Results included the entire treated group (ie, all selected cancers) and groups stratified by cancer type.

Table 2.

Age Sensitivity Analysis for All Exposure Group Analyses

To examine differences among individual statins (Fig 4), we stratified molecules by natural and synthetic origin. Compared with simvastatin as a reference, the more hydrophilic rosuvastatin (HR, 1.21; 95% CI, 1.03 to 1.43) and fluvastatin (HR, 2.18; 95% CI, 1.36 to 3.50) molecules were associated with a higher rate of death, but no differences were observed between other products, (ie, simvastatin was associated with a protective effect). When grouped by type, natural statins had a marginally protective effect on survival, but this was not statistically significant (HR, 0.91; 95% CI, 0.83 to 1.01)

Fig 4.

The effect of statin stratification by type and class on 1-year survival. In comparison by type, simvastatin was used as reference (Ref). In comparison by class, synthetic statins were used as reference (Ref). The list of natural statins were lovastatin, pravastatin, and simvastatin. Synthetic statins were rosuvastatin, atorvastatin, fluvastatin, and pitavastatin. The full multivariable Cox model included the following adjustment covariates: age; Charlson Comorbidity Index score; region; uses of anticoagulants, antihypertensives, antiplatelets, antiarrhythmics, or digoxin; Elixhauser Index comorbidities; and pre-Index history of coronary heart disease, deep vein thrombosis, pulmonary embolism, atrial thrombosis, or myocardial infarction. Hazard ratios (HRs) and associated 95% CIs for the effect of treatment group are presented.

Bisphosphonate users (n = 4,528) were compared with nonusers and stratified by statin use. Among all cancers, bisphosphonates were associated with a nonsignificant reduction in death when compared with nonusers (HR, 0.82; 95% CI, 0.65 to 1.03; Fig 5). Stratification by cancer type could not be completed because of the limited population size. A treatment group that consisted of patients who received both statins and bisphosphonates (n = 4,090) exhibited a much larger improvement in survival compared with a subset of nonusers who did not receive either medication (HR, 0.60; 95% CI, 0.45 to 0.81).

Fig 5.

The effect of statin combinations on 1-year survival. Bisphosphonates assessed in the analysis were alendronate sodium, etidronate disodium, ibandronate sodium, pamidronate disodium, risedronate sodium, and zoledronic acid. Bisphosphonate and dipyridamole users were required to have at least 90 cumulative days supplied in the 6 months before cancer diagnosis. The full multivariable Cox model included the following adjustment covariates: age; Charlson Comorbidity Index score; region; uses of anticoagulants, antihypertensives, antiplatelets, antiarrhythmics, or digoxin; Elixhauser Index comorbidities; and pre-Index history of coronary heart disease, deep vein thrombosis, pulmonary embolism, atrial thrombosis, or myocardial infarction. Hazard ratios (HRs) and associated 95% CIs for the effect of treatment group are presented.

The majority of patients remained on statin therapy; an average of 78.3% of days in the postindex period were covered by statin therapy, and only 19 of 65,440 patients stopped statin therapy after cancer diagnosis. More than half of this treatment group received statins for the entire follow-up period. Generally, we observed that the statin cohort continued to receive statin therapy for the majority of their postdiagnosis follow-up times.

DISCUSSION

This observational cohort study used epidemiologic data and identified a significant effect on the overall survival of patients with cancer who receive statins at the time of diagnosis. This advantage was specific to cancers of the lung and kidney and to leukemias. In addition, the concurrent use of a bisphosphonate with a statin was associated with an improvement in overall survival, but stratification by cancer type was not possible because of the small sample size. This observation is consistent with the mechanisms of action of the two agents, which suggests the theoretical potential for synergy in a prospective study. The effect of dipyridamole in combination with statin was not significant in the small cohort of patients who received this combination (n = 651). The effect of statins was comparable to that of other cholesterol-lowering medications, which is consistent with a recent cohort study in postmenopausal women that showed that regular use of statins or other lipid-lowering medications was associated with decreased cancer death.³ Multiple systems or pathways, including inhibition of Ras, improvement of immune surveillance, and the reduction of venous thromboembolisms (VTEs), may explain the mechanism by which statins induce their effect on cancer survival.

Statins affect normal cell survival mechanisms, which include cell proliferation, proapoptotic effects, induction of autophagy, and anti-invasive and antimigration effects that have been systematically studied in both in vitro and in vivo model systems. The overarching hypothesis involves the potential effect of statins on the Ras signaling pathway.^55-61 Lung cancer, pancreatic cancer, and hematopoietic/lymphoid cancers are associated with high rates of KRAS mutation (38%, 63%, and 21%, respectively).⁶² High frequency of mutation may explain why patients with these cancers benefited in our analysis. However, our results did not show an effect in colorectal cancer, which also has a high frequency of KRAS mutation.⁶² Statin use after diagnosis was shown previously to reduce colorectal cancer–specific mortality,¹¹ and the prospective use of simvastatin with cetuximab plus irinotecan in patients with KRAS mutations had a favorable disease control rate.²¹ When in vitro evidence is considered, statin-mediated modulation of protein prenylation in cancer cells requires suprapharmacologic concentrations (eg, 1 to 25 µM) for prolonged periods.^63-66 Previous studies from our group demonstrated that high simvastatin doses (7.5 mg/kg twice daily), approximately 25-fold of the typical daily dose (ie, 40 mg), achieved maximum plasma concentrations that were in the range of 0.08 to 2.2 µM.⁶⁷ Consistent with this, our results show that the dose intensity of typical dosages of cholesterol-lowering treatment do not affect overall survival, which suggests that other mechanisms may be involved.

Prior studies have suggested that the immunostimulatory effects of statins are the mechanism for anticancer activity. Through inhibition of the mevalonate pathway, statins can induce innate lymphocyte activation and increase immune surveillance. Depletion of prenyl pyrophosphates in human dendritic cells generates danger signals that can translate into caspase-1 activation. Caspase-1 cleaves interleukin-1-beta (IL-1β) and IL-18 into their activated forms, which allows the release of cytokines that include interferon gamma and IL-2.⁶⁸ Statin-induced activation of IL-2–primed γδ T cells and natural killer cells causes potent antitumor cytotoxicity, and ectopic GGPP reintroduced into the cell culture abolishes this effect.^69,70 Immunomodulation may be an important contributor, because immune cells are more likely than cancer cells to be exposed to higher statin concentrations.

Finally, given the lack of difference between statin and nonstatin drugs, the survival effect may in part be due to reductions in thrombotic events. Observational studies show that statins lower the risk of VTEs in the cancer population, which thereby increases overall survival.⁷¹ In a prospective observational cohort of 1,434 patients with cancer, VTEs occurred in 2.94% of patients at 12 months and in 3.54% at 24 months for statin users. In comparison, those who were not treated with statins had elevated rates of VTE of 7.13% at 12 months and 8.13% at 24 months (P = .04). Among newly diagnosed patients with cancer who were observed prospectively for a 2-year period, statins users had a lower risk of VTE than nonusers (HR, 0.43; 95% CI, 0.19 to 0.98). In contrast, a meta-analysis of 27 trials to assess the effect of statin use and the lowering of LDL cholesterol on cancer incidence and mortality found a lack of effect after a median of 5 years of therapy but a small effect within the first year after diagnosis.⁷² In the Algorithm for Comorbidities, Associations, Length of Stay and Mortality (ACALM) study, hyperlipidemia was associated with a significantly reduced mortality rate in lung, breast, prostate, and bowel cancers.⁷³ With respect to age, our analyses shows that those older than 65 years had no significant benefit from statin treatment, which may be the result of an overall increased VTE risk in that population. More work is needed to understand how cholesterol-lowering medications affect the risk of thrombotic events in cancer and whether this effect explains part of the overall protective effect observed in this study.

This study is subject to the limitations of all claims-based studies.^74,75 Claims data lack detailed information on laboratory values or tumor staging, which may have influenced the outcomes of this study. This study was limited to 1 year of follow-up because of the availability of data and the heterogeneity and confounded time variance associated with longer follow-up. Last, although propensity-score matching is known to reduce selection bias in nonrandomized studies, it is possible that residual bias was present or that unmeasured confounders may have affected these findings.⁵⁰ This study is strengthened by a large sample size, by inclusion of minimum medication exposure criteria (eg, at least 90 cumulative days of medication supplied in the 6 months before diagnosis), and by inclusion of an active control group; factors lacking in similar studies.

In conclusion, epidemiologic health outcomes data can be used to test hypotheses that are based on the effect of drugs on specific biologic pathways and processes. Our work shows that the use of statins alone and in combination with bisphosphonates could provide a survival benefit in certain cancers. Additional prospective clinical studies are warranted.

AUTHOR CONTRIBUTIONS

Conception and design: Sherif El-Refai, Susanne M. Arnold, Esther P. Black, Markos Leggas, Jeffery C. Talbert

Administrative support: Esther P. Black, Markos Leggas

Financial support: Esther P. Black, Markos Leggas

Provision of study material or patients: Susanne M. Arnold, Markos Leggas

Collection and assembly of data: Sherif El-Refai, Joshua D. Brown, Susanne M. Arnold, Markos Leggas

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Epidemiologic Analysis Along the Mevalonate Pathway Reveals Improved Cancer Survival in Patients Who Receive Statins Alone and in Combination With Bisphosphonates

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/site/ifc.

Sherif M. El-Refai

No relationship to disclose

Joshua D. Brown

No relationship to disclose

Susanne M. Arnold

Research Funding: AstraZeneca (Inst), Bristol Myers Squibb Oncology (Inst), Cancer Research and Biostatistics (Inst), Amgen (Inst), Stemcentrix (Inst), Genentech (Inst)

Esther P. Black

No relationship to disclose

Markos Leggas

No relationship to disclose

Jeffery C. Talbert

No relationship to disclose