Mortality among US veterans with a physician-documented diagnosis of pyruvate kinase deficiency

ABSTRACT

Real-world studies of pyruvate kinase (PK) deficiency and estimates of mortality are lacking. This retrospective observational study aimed to identify patients with PK deficiency and compare their overall survival (OS) to that of a matched cohort without PK deficiency. Patients with ≥1 diagnosis code related to PK deficiency were selected from the US Veterans Health Administration (VHA) database (01/1995–07/2019); patients with a physician-documented diagnosis were included (PK deficiency cohort; index: date of first diagnosis code related to PK deficiency). Patients in the PK deficiency cohort were matched 1:5 to patients from the general VHA population (non-PK deficiency cohort; index: random visit date during match’s index year). OS from index was compared between the two cohorts. Eighteen patients in the PK deficiency cohort were matched to 90 individuals in the non-PK deficiency cohort (both cohorts: mean age 57 years, 94% males; median follow-up 6.0 and 8.0 years, respectively). At follow-up, patients in the non-PK deficiency cohort had significantly longer OS than the PK deficiency cohort (median OS: 17.1 vs. 10.9 years; hazard ratio: 2.3; p = 0.0306). During their first-year post-index, 75% and 40% of the PK deficiency cohort had laboratory-confirmed anemia and iron overload, respectively. Among patients who died, cause of death was highly heterogeneous. These results highlight the increased risk of mortality and substantial clinical burden among patients with PK deficiency. While the intrinsic characteristics of the VHA database may limit the generalizability of the results, this is the first real-world study to characterize mortality in patients with PK deficiency.

KEYWORDS:

• anemia
• hemolytic
• mortality
• pyruvate kinase deficiency
• Veterans Health Administration

Introduction

Pyruvate kinase (PK) deficiency is a rare, inherited disorder caused by autosomal recessive mutations in the PKLR gene, whereby a glycolytic defect causes reduced adenosine triphosphate levels and leads to hemolytic anemia [1–3]. Although its prevalence in the United States (US) is unknown, clinically diagnosed PK deficiency is likely to affect between 3.2 and 8.5 individuals per million in Western populations [4]. To date, more than 300 mutations have been associated with PK deficiency [5]. This molecular heterogeneity is in turn likely to contribute to the highly variable clinical presentation of the disease [6]. Notably, symptomatic anemia may require lifelong red blood cell (RBC) transfusions among certain patients with PK deficiency [7,8]. A myriad of complications related to chronic hemolytic anemia and its current treatment may also arise, including osteoporosis, pulmonary hypertension, iron overload, and liver cirrhosis [9]^.

Until the recent US Food and Drug Administration approval of mitapivat [10], a disease modifying, first-in-class, oral allosteric activator of red-cell PK, for the treatment of hemolytic anemia in adults with PK deficiency, and its approval by the European Union Medicines Agency and the Medicines and Healthcare products Regulatory Agency in Great Britain for the treatment of PK deficiency in adults [11,12], the standard of care for PK deficiency included measures that were supportive only and non-disease-specific, such as RBC transfusions, treatment and prevention of iron overload, and removal of the spleen or gallbladder [3,13]. Despite their potential benefits, supportive therapies such as transfusion regimens and/or splenectomy are also associated with long-term complications, such as iron overload, thrombosis, or post-splenectomy infection, which could further add to the burden of disease [14].

Historically, the identification of patients with PK deficiency in real-world data has been challenging due to a lack of diagnosis codes and treatments specific to this disorder. As a result, population-based studies of PK deficiency using administrative claims or electronic medical records databases are currently lacking. Moreover, outcomes data on mortality in this patient population remain limited to a small number of individual case reports [14–22].

Improved identification of patients with PK deficiency in real-world data and a better understanding of their survival outcomes would help to provide insight into the burden of disease among this population. Accordingly, the present study aimed to identify patients with a diagnosis of PK deficiency in the US as documented by physicians, to compare their overall survival to an age- and gender-matched cohort of individuals without PK deficiency, and to explore the causes of death for patients with PK deficiency.

Methods

Data source

This retrospective observational study was conducted using electronic medical records and physicians’ notes from the US Veterans Health Administration (VHA) [23]. The US VHA is the largest integrated health care system in the US, providing a set of comprehensive services to veterans at 1,255 health care facilities (including 170 Veterans Affairs [VA] medical centers and 1,074 outpatient clinics) to over 9 million veterans enrolled in the VA health care program. It includes information regarding demographics, vital signs, laboratory results, diagnoses, procedures, inpatient and outpatient services, drug prescriptions, and database enrollment history. It also includes physician notes that are part of patients’ medical records, such as progress and consult notes from primary care, specialty care, mental health, social work, and nursing notes. Although patients in the national VHA population are predominantly male, the database was selected for this research due to its long length of follow-up, availability of physician notes to confirm PK deficiency diagnoses, and availability of death data.

All data used in the present study were de-identified in compliance with the Health Insurance Portability and Accountability Act (HIPAA) and in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments. The study protocol was approved by the Institutional Review Board (IRB) and the Research and Development Committee (RDC) of the Southeast Louisiana Veterans Health Care Systems (IRB approval number: 629-652).

Sample selection and study design

Two cohorts were defined for this analysis, including a PK deficiency cohort and a matched non-PK deficiency cohort. For the PK deficiency cohort, patients with ≥1 diagnosis code related to PK deficiency between January 1995 and July 2019 were identified from the US VHA database. At the time of analysis, there was no diagnosis code specific to PK deficiency. Therefore, the following diagnosis codes from the International Classification of Disease, Ninth and Tenth Revisions, Clinical Modification (ICD-9-CM and ICD-10-CM, respectively) were used for identification: anemia due to disorders of glycolytic enzymes (ICD-10-CM D55.2), other hemolytic anemias due to enzyme deficiency (ICD-9-CM 282.3), or unspecified hereditary hemolytic anemia (ICD-9-CM 282.9, ICD-10-CM D58.9). To be considered for inclusion, physicians’ notes were required to contain all the three keywords ‘pyruvate’, ‘kinase’, and ‘deficiency.’ Finally, a manual review of these physicians’ notes was performed to identify patients with a physician-documented diagnosis of PK deficiency; no confirmation via enzymatic activity or genetic testing was available. The index date for the PK deficiency cohort was defined as the date of the first medical record with a diagnosis code related to PK deficiency. Each patient in the PK deficiency cohort was matched 1:5 by age at index, sex, and index year (±1 year) to patients from the general VHA population with no diagnosis codes related to PK deficiency. The index date for the non-PK deficiency cohort was defined as a random visit date during their match’s index year. The time periods for this study included a baseline period, follow-up period, and study period. The baseline period was defined as the 12-month period prior to the index date. The follow-up period was defined as the 12-month period after the index date. Finally, the study period was defined as the time period from the index date to the date of death from any cause or the end of data.

Measures and outcomes

Demographics and clinical characteristics were summarized for the PK deficiency cohort and their non-PK deficiency cohort matches. Demographic information recorded at the index date included age, sex, race, region, weight, height, body mass index (BMI), and year of index date. Charlson Comorbidity Index (CCI) score was summarized for the 12-month baseline period prior to index date [24].

Laboratory markers, including hemoglobin, reticulocytes, lactate dehydrogenase (LDH), and serum ferritin, were summarized only for the PK deficiency cohort during the 12-month follow-up period and study period. Clinical characteristics were summarized for both cohorts during the 12-month follow-up period. Overall survival (OS) from the index date was summarized over the study period for the PK deficiency cohort and their non-PK deficiency cohort matches. Causes of death among patients in the PK deficiency cohort who died were identified from the National Death Index (NDI).

Statistical analyses

Patient demographic and clinical characteristics at baseline were summarized between the PK deficiency cohort and the non-PK deficiency cohort after matching. Means, medians, standard deviations, and ranges were calculated for continuous variables, while frequency counts and percentages were calculated for categorical variables. The number of patients with available laboratory results during the follow-up period and the study period was summarized, along with median values and range of laboratory results, as well as the number and proportion of patients that met pre-determined criteria (e.g. average hemoglobin concentration of <12 g/dL, maximum serum ferritin concentration >1000 ng/mL, average serum ferritin concentration >500 ng/mL).

OS from the index date was summarized for each cohort using the Kaplan-Meier method and compared between cohorts using a univariate Cox proportional hazards model with robust standard error estimation. A Schoenfeld’s global test was used to check the proportional hazards assumption. Patients were censored at the date of end of data availability. Causes of death among patients in the PK deficiency cohort who died were summarized based on NDI Underlying Cause ICD-10-CM.

Results

Patient characteristics

A total of 18 patients met inclusion criteria for the PK deficiency cohort and were matched to 90 individuals in the non-PK deficiency cohort. Baseline characteristics for both cohorts are shown in Table 1. For both cohorts, the mean age at index was 57 years, and 94% of patients were male; 83–85% were white and the mean CCI score was 0.4–0.5, with no significant differences between cohorts in terms of these characteristics. Imbalances remained for region, with more patients from the South in the PK deficiency cohort, and for BMI, with higher mean BMI in the non-PK deficiency cohort.

Table 1. Patient characteristics.

	PK deficiency cohort N = 18	Matched non-PK deficiency cohort N = 90
Age at index year, mean ± SD [median]	56.8 ± 13.6 [59.0]	56.8 ± 13.1 [59.0]
Category, n (%)
21–30 years	1 (5.6)	5 (5.6)
31–40 years	0 (0.0)	0 (0.0)
41–50 years	5 (27.8)	25 (27.8)
51–60 years	4 (22.2)	20 (22.2)
61–70 years	7 (38.9)	35 (38.9)
71–80 years	0 (0.0)	0 (0.0)
81+ years	1 (5.6)	5 (5.6)
Male, n (%)	17 (94.4)	85 (94.4)
Region, n (%)
South	6 (33.3)	6 (6.7)
Midwest	2 (11.1)	4 (4.4)
North East	7 (38.9)	48 (53.3)
West	3 (16.7)	32 (35.6)
White, n (%)	15 (83.3)	77 (85.6)
Weight (lbs), mean ± SD [median]	190.0 ± 42.4 [189.2]	206.2 ± 47.5 [195.0]
Height (in), mean ± SD [median]	69.3 ± 2.8 [69.9]	69.0 ± 2.9 [69.9]
BMI, mean ± SD [median]	27.6 ± 5.0 [26.7]	30.5 ± 6.7 [28.8]
20–25, n (%)	7 (38.9)	11 (12.2)
26–30, n (%)	5 (27.8)	37 (41.1)
>30, n (%)	6 (33.3)	39 (43.3)
Charlson Comorbidity Index score, mean ± SD [median]	0.4 ± 1.2 [0]	0.5 ± 1.2 [0]

Abbreviations: PK = pyruvate kinase; SD = standard deviation.

Laboratory markers of anemia, hemolysis, and iron overload for the PK deficiency cohort (Table 2)

In the PK deficiency cohort, 16/18 patients had at least one laboratory test documenting hemoglobin during the first year post-index, and 18/18 during the full study period. The median hemoglobin among these patients was 11.03 g/dL (range: 6.30–16.82 g/dL) during the first year post-index. In total, 12/16 patients had an average hemoglobin <10 g/dL during the first year post-index, and 13/18 patients had an average hemoglobin <12 g/dL during the full study period.

Table 2. Laboratory findings for 1-year post-index and full follow-up period for PK deficiency cohort (N = 18).

	Hemoglobin (g/dL)		Serum ferritin (ng/mL)		Reticulocytes (%)		LDH (u/L)
	1-year post-index	Full follow-up period	1-year post-index	Full follow-up period	1-year post-index	Full follow-up period	1-year post-index	Full follow-up period
Number with ≥1 test	16	18	10	14	11	14	11	14
Mean	10.88	10.73	643.73	772.48	6.13	6.97	432.33	382.19
SD	2.70	2.55	401.04	529.38	4.88	5.65	295.73	282.10
Median	11.03	10.30	638.33	517.21	4.28	4.74	230.00	224.00
Range (min, max)	6.30, 16.82	6.30, 16.82	65.50, 1170.00	57.78, 1669.44	2.28, 19.00	1.88, 23.75	187.00, 999.58	108.00, 985.13

Abbreviations: g/dL = grams per deciliter, LDH = lactate dehydrogenase, ng/mL = nanograms per milliliter, SD = standard deviation, u/L = units per liter.

There were 11/18 patients in the PK deficiency cohort with at least one lab test documenting the percentage of reticulocytes during their first year post-index, and 14/18 during the full study period. Among these patients, the median percentage of reticulocytes was 4.3% (range: 2.3%–19.0%) during the first year post-index. Furthermore, 11/18 patients in the PK deficiency cohort had at least one lab test documenting LDH during their first year post-index, and 14/18 during the full study period. The median LDH among these patients was 230.0 u/l (range: 187.0–999.6 u/l) during the first year post-index, and 8/11 patients had at least one test that was higher than the normal range.

There were 10/18 patients in the PK deficiency cohort with at least one lab test documenting serum ferritin during their first year post-index, and 14/18 during the full study period. The median serum ferritin among these patients was 638.33 ng/mL (range: 65.50–1170.00 ng/mL) during the first year post-index. Maximum serum ferritin >1,000 ng/ml was observed among 4/10 patients during their first year post-index, and 6/14 patients during the full study period. Moreover, an average serum ferritin >500 ng/ml was observed among 5/10 patients during their first year post-index, and 7/14 patients during the full study period.

Clinical characteristics

Four patients in the PK deficiency cohort had a diagnosis code of osteoporosis after the index date compared to 5 patients in the non-PK deficiency cohort (22.2% vs. 5.6%; p = 0.0406). Two patients in the PK deficiency cohort had a diagnosis of cirrhosis in the 1-year period post index compared to one patient in the non-PK deficiency cohort (11.1% vs. 1.1%; p = 0.0714).

Three patients in the PK deficiency cohort had at least one red blood cell transfusion in the 1-year period after the index date compared to none in the non-PK deficiency cohort (16.7% vs. 0.0%; p = 0.0040). One patient in the PK deficiency cohort had a splenectomy in the 1-year period post index compared to none in the non-PK deficiency cohort (5.6% vs. 0.0%; p = 0.1667). One cholecystectomy was observed in each group post index date (5.6% vs. 1.1%; p = 0.1889). In patients with PK deficiency, cholecystectomy and splenectomy are often performed at the same time, typically during childhood. Because the VHA database only captures procedures that occurred while patients were enrolled in the VHA health system, it is not feasible for us to report the actual percentage of patients who have had a splenectomy or cholecystectomy during their lifetime.

Overall survival

The median follow-up was 6.0 years for the PK deficiency cohort (mean ± standard deviation [SD] = 7.3 ± 5.2) and 8.0 years for the non-PK deficiency cohort (9.2 ± 5.8). Over the follow-up period, there were 9 (50%) deaths in the PK deficiency cohort and 28 (31%) deaths in the non-PK deficiency cohort. The median time until death was 10.9 years for the PK deficiency cohort and 17.1 years for the non-PK deficiency cohort; the Kaplan–Meier curves for both cohorts are shown in Figure 1. Patients in the PK deficiency cohort had a significantly shorter OS than the non-PK deficiency cohort (Hazard Ratio [HR]: 2.3; p = 0.0306); there was no evidence of violation of the proportional hazards assumption. At 10 years after the index date, the probability of mortality was 42% (95% confidence interval [CI]: 21%–72%) in the PK deficiency cohort compared with 28% (19%–41%) in the non-PK deficiency cohort.

Figure 1. Kaplan–Meier survival estimates for PK deficiency cohort and matched non-PK deficiency cohort. Abbreviations: PK = pyruvate kinase.

Causes of death

In the PK deficiency cohort, there were a total of nine documented deaths during the study period. The median age at death was 70 years. The cause of death included two deaths related to liver disease, two due to heart disease, two due to acute myeloid leukemia (AML), and two due to lung cancer; one death was related to dementia.

Discussion

To the knowledge of the authors, this is the first retrospective, real-world study of patients with PK deficiency using administrative claims or electronic medical records. Moreover, this is the first study evaluating mortality in a group of patients with PK deficiency outside of the case report literature. Our results provide evidence of a higher 10-year probability of mortality (42% [95% CI: 21%–72%] vs. 28% [19%–41%]) and significantly shorter median OS (10.9 years vs. 17.1 years; HR: 2.3; p = 0.0306) among patients with PK deficiency in our study sample compared to their matched controls without PK deficiency. Cause of death among patients with PK deficiency was highly variable and included potential complications (e.g. liver disease) as well as comorbid conditions (e.g. heart disease, lung cancer, and dementia). Patients with PK deficiency showed evidence of hemolytic anemia based on laboratory markers, including low hemoglobin in combination with high levels of reticulocytes and LDH. During the first year post-index, the median hemoglobin level was ∼11.0 g/dL and about 75% of patients had hemoglobin <12 g/dL; meanwhile, the median percentage of reticulocytes was higher than normal levels (4.3% vs. 0.5–2.5%[19,25]) and median LDH was in the upper range of normal (230.0 u/l vs. 140–280 u/l [26]). Given the elevated levels of reticulocytes and LDH observed among patients the present study, it is likely that this anemia was secondary to hemolysis. Evidence of iron overload (a maximum serum ferritin concentration >1000 ng/mL or an average serum ferritin concentration >500 ng/mL [8]) was observed in up to half of all patients with available serum ferritin data. Although iron overload is recognized as a common complication of PK deficiency regardless of RBC transfusion status, as well as a major contributor to morbidity and mortality, not all physicians are aware [15,19–21,27]. Taken together, the increased risk of mortality and tendency toward anemia and iron overload among patients in our study highlights the substantial burden of disease among patients with PK deficiency, the need for accurate follow-up to prevent complications, and possibly the need for safe and effective disease-specific treatments for this population.

To date, ten case studies have documented early mortality among 12 patients with PK deficiency from different geographic regions, including five recorded deaths among infants and seven deaths among patients ranging between 17 and 45 years of age [14–22]. Most of these patients had a complex history of medical interventions, complications, and comorbidities leading up to their eventual death. Cause of death varied widely among these patients. Notably, deaths in infancy due to progressive liver failure [17] and complications following exchange transfusion [21] have been reported. Among patients who succumbed later in life, one patient died from complications following HSCT at the age of 17 [14], three patients died due to septic shock at the ages of 24, 41, and 43 [16,22,27], and four patients aged 39–45 years died from complications related to iron overload (e.g. heart failure) [15,19,20,28].

It is possible that certain deaths in the present study (i.e. deaths from liver disease and possibly heart disease) could have been complications of PK deficiency, whereas deaths due to progressive diseases such as lung cancer or dementia may reflect comorbidities related to the advanced age of the US VHA study sample. An important caveat is that we could not confirm the extent to which a death might have been related to the complications of PK deficiency, as details from port-mortem analyses were not available. Additionally, AML was a cause of two deaths among patients with PK deficiency in the present study, which was unexpected given the rarity of AML in the general population. One hypothesis regarding the deaths due to AML is that these patients might have had acquired PK deficiency, rather than an inherited form due to mutations in the PKLR gene. Acquired PK deficiency in patients with clonal myeloid neoplasms, including AML, has been described multiple times previously [29–31].

The results of the present study should be viewed within the context of certain additional limitations. First, a small sample size of patients with PK deficiency was captured in our study compared to what might have been expected given the disease prevalence, as it is possible that some patients with PK deficiency avoided or were not eligible for military service. In that regard, the PK deficiency population in our study may be skewed toward individuals who ostensibly had milder disease at the time of their enlistment and had not yet been diagnosed with any disqualifying conditions or were not actively manifesting signs or symptoms that would have rendered them ineligible. Indeed, although patients with PK deficiency in our sample had shorter OS relative to matched controls, the median age of death in the PK deficiency group was 70 years, which is considerably older than the age of death in prior case reports. Second, in our analysis, the matching of patients with and without physician-documented PK deficiency helped to minimize confounding due to differences in observed baseline characteristics between the two cohorts. However, due to the retrospective observational nature of this study, the analyses may have been affected by unobserved factors that could not be controlled for through the matching process. Another limitation of this study was the lack of available genotype information, which could have helped confirm the PK deficiency diagnosis and distinguish patients with acquired rather than inherited PK deficiency. We acknowledge the possibility of misdiagnosis; this concern applies to similar research in many other diseases, especially for rare diseases and studies using anonymized, retrospective data sources. A further limitation of this study was that our US VHA sample consisted of mostly older males. Consequently, the younger (i.e. pediatric and adolescent) and female populations with PK deficiency are underrepresented, which may limit the generalization of our findings beyond the study sample. Though not a demographically representative population, this US VHA database is one of very few data sources with the longitudinality, level of clinical detail, and mortality data needed for such an analysis. Additionally, our study considered a wide period of observation (1995–2019) with the purpose of maximizing the sample size of patients with PK deficiency, given the rarity of the disease; these results should be interpreted in the context of the time period assessed, as there may be differences in disease identification and management in more recent years. Finally, there was no specific ICD-10-CM code for PK deficiency at the time of this analysis; therefore, umbrella codes were used as a first step to identify potential patients with PK deficiency. As of October 2021, a new PK deficiency-specific ICD-10-CM code has been made available, which may help to facilitate future research regarding the burden of disease among patients with this rare, debilitating condition.

In conclusion, the results of this retrospective study are the first of its kind to suggest that patients with PK deficiency may be at an increased risk of mortality compared with the general population. Further research aimed at understanding the cause of death in this population is warranted, as is the replication of this study using larger sample sizes and other real-world data sources that better represent females and the pediatric and adolescent PK deficiency age groups.

Funding details

This work was funded by Agios Pharmaceuticals, Inc.

Acknowledgments

Medical writing and editorial assistance was provided by Mona Lisa Chanda, PhD, an employee of Analysis Group, Inc. Support for this assistance was provided by Agios Pharmaceuticals, Inc.

Disclosure statement

Erin Zagadailov was an employee and stockholder of Agios Pharmaceuticals, Inc. at the time of the development of this project, and an employee of Takeda Oncology within the 3 years prior to the development of this project. Audra N. Boscoe and Bryan McGee are employees and stockholders of Agios Pharmaceuticals, Inc. Hanny Al-Samkari was a consultant for Agios, argenx, Sobi, Moderna, Novartis, Rigel, Forma and received research funding from Agios, Amgen, Sobi. Viviana Garcia-Horton and Sherry Shi are employed by Analysis Group, Inc., which received payment from Agios Pharmaceuticals, Inc. for participation in this research. Lizheng Shi has a research contract and received consulting fees from Analysis Group, Inc. Dendy Macaulay was an employee of Analysis Group, Inc. at the time of this project.

Data availability statement

The dataset supporting the conclusions in this article is available from the US VHA, which we had full permissions to access and use for this study. However, restrictions apply to the availability of these data, which were used under license for the current study, and are therefore not publicly available.

Additional information

Notes on contributors

Erin Zagadailov

Erin Zagadailov is a former Director of Health Economics & Outcomes Research (HEOR) at Agios Pharmaceuticals and is a biotech executive with experience in patient-focused drug development, real-world evidence, and health economics. She is currently the Chief Innovation Officer at Clinical Outcomes Research Group, LLC. Her research has been published in The New England Journal of Medicine, The Lancet, and the British Journal of Hematology, among other major peer-reviewed journals.

Hanny Al-Samkari

Hanny Al-Samkari is an Assistant Professor of Medicine at Harvard Medical School as well as an NIH-funded clinical investigator and classical hematologist at the Massachusetts General Hospital (MGH) Division of Hematology Oncology. He also serves as Co-Director the MGH Hereditary Hemorrhagic Telangiectasia (HHT) Center of Excellence. His clinical and research interests are in hemostasis, thrombosis and hemolysis, with focuses in HHT, immune thrombocytopenia, and hereditary hemolytic anemias. He currently serves as PI for many clinical trials in these areas and is a recognized expert in the clinical development of novel therapeutics for HHT, hemolytic anemias and immune thrombocytopenia. He is the current Executive Editor of Hematology: The ASH Education Program, an American Society of Hematology peer-reviewed publication, and has published over 130 peer-reviewed manuscripts. His original research has been featured in The New England Journal of Medicine, The Lancet, and Blood, among other major peer-reviewed journals in hematology and medicine.

Audra N. Boscoe

Audra N. Boscoe leads the Health Economics, Outcomes Research & Data Science Analytics department at Agios Pharmaceuticals. Her areas of specialty include real-world evidence generation using a variety of data sources (disease registries, surveys, electronic medical records, and administrative claims databases), economic modeling, and patient-reported outcomes. Her career has spanned a variety of therapeutic areas including oncology, hematology, cardiology, nephrology, and neurology, with a particular focus on rare diseases. Dr. Boscoe has published numerous articles in peer-reviewed journals and presented her original research at a number of clinical and health economics and outcomes research conferences.

Bryan McGee

Bryan McGee is a Sr. Medical Director for the PK deficiency program at Agios Pharmaceuticals, Inc. He currently serves as the medical affairs lead for the PK deficiency Peak registry, is a member of the Peak Registry Steering Committee, and has co-authored multiple abstract and manuscript submissions on the burden of disease in PK deficiency. His career has spanned a variety of therapeutic areas including hematology, rheumatology, endocrinology, and pediatric/adult infectious disease with a particular focus on rare diseases.

Sherry Shi

Sherry Shi is a biostatistician who specializes in health economics and outcomes research (HEOR), biostatistics, and epidemiology. Her experience includes retrospective analyses of various data sources such as medical insurance claims and electronic medical records. Sherry has conducted data analysis in multiple clinical areas, such as oncology, hematology, and gastroenterology. Her research has been published in numerous peer-reviewed medical journals, including the Journal of Medical Economics (JME), the Journal of Comparative Effectiveness Research, Journal of Cancer Therapy. Prior to joining Analysis Group, Sherry was a research assistant at McGill University.

Dendy Macaulay

Dendy Macaulay is a former Vice President of Analysis Group and is an economist with extensive experience in the application of statistics and econometrics to health outcomes research. Her expertise includes retrospective claims analyses, patient chart reviews, clinical trial data analyses, economic modeling, and health economics and outcomes research strategy. Her work has involved the development economic models to evaluate Medicare demonstration projects and drug expansion into emerging markets; comparative effectiveness and cost effectiveness research; prediction algorithms; the assessment of disease prevalence, burden of illness, and treatment patterns; and the development pharmaceutical pricing strategies. Dr. Macaulay's experience encompasses a wide range of therapeutic areas, including oncology, mental health, respiratory, immunology, and other chronic disease areas with the use of multiple data sources, including clinical trial, Medicare, Medicaid, private insurance, electronic medical record, and patient chart review data. Her experience also includes analyses in support of health care commercial litigation related to patents and pharmaceutical licensing. Dr. Macaulay has published numerous articles in peer-reviewed journals, including Health Affairs, Respiratory Medicine, PharmacoEconomics, and The Journal of the American Geriatrics Society.

Lizheng Shi

Lizheng Shi is an Endowed Regents Professor and Interim Chair in the Department of Global Health Policy and Management in the Tulane University School of Public Health and Tropical Medicine (TUSPHTM). He is also the director for the Health Systems Analytics Research Center (HSARC) and a member of the editorial board for Pharmacoeconomics. Dr. Shi also has a joint appointment with the Department of Medicine at the Tulane University School of Medicine. His research interests include health technology assessments and health care quality, access, and economics. Prior to his appointment at Tulane, Dr. Shi held the position of Senior Health Outcomes Scientist at Eli Lilly and Company in Indianapolis.

Viviana Garcia-Horton

Viviana Garcia-Horton is a statistician who specializes in the application of advanced statistical techniques to the field of health economics and outcomes research (HEOR). She has broad experience developing research and strategies in a wide range of disease areas, including diabetes, oncology, hematology, rare diseases, dermatology, psychiatry, neurology, and gastroenterology. Dr. Garcia-Horton’s expertise includes causal inference for experimental and observational data, predictive modeling, missing data problems, Bayesian statistics, clinical trial design and analysis, indirect treatment comparisons, individualized medicine, and retrospective claims database analyses. Her work includes developing network meta-analyses for regulatory submission, treatment patterns and economic outcomes research using retrospective claims analyses, clinical trial data analyses, and analyses of patient-reported outcomes (PROs) to support approved product labeling. Dr. García-Horton’s research has been published in peer-reviewed journals and presented at numerous clinical research, health economics, and statistics conferences.