Dr D. M. Evon, Division of Gastroenterology and Hepatology, CB# 7584, 8010 Burnett-Womack, University of North Carolina, Chapel Hill, NC 27599, USA. E-mail: email@example.com
Background Most research on the psychiatric symptoms of peginterferon/ribavirin for the treatment of hepatitis C comes from VA centres and clinical trials with rigid entry criteria that often excluded patients with markers of mental health and substance use disturbance (MH/SUD). The findings from these lines of research may not be generalizable to patients treated under standard of care in a tertiary care setting.
Aim To investigate the incidence and outcomes of psychiatric symptoms in patients treated under standard of care protocol, not enrolled in clinical trials.
Methods This is a retrospective analysis of 215 patients who underwent therapy from 2002 to 2006 at a university-based tertiary care centre. Survival curves explored the relationship between history of MH/SUD and the development of psychiatric symptoms on treatment.
Results The cumulative history of MH/SUD was 67%. Of these, 39% had taken psychotropic medications previously, and 80% continued on them during therapy. On therapy, 46% developed depressive symptoms, 19% and 46% endorsed anxiety and irritability respectively. Cumulatively, 64% of patients indicated mood disturbance on therapy. Most symptoms developed between weeks 2 and 18, and rarely after week 20. Of those who developed mood symptoms, 66% required an intervention. Treatment discontinuation was infrequent.
Conclusions This large observational study provides important insights into the incidence and course of psychiatric symptoms in an unselected sample of patients treated in a tertiary care setting. Patients had higher rates of MH/SUD comorbidity, psychotropic medication use and exhibit higher rates of mood disturbance on therapy compared with previous reports, although a majority completed the prescribed treatment regimen.
Combination therapy with peginterferon and ribavirin (PegIFN/RBV) is highly effective for the treatment of chronic hepatitis C and yields a sustained virological response in 54–56% of those treated.1–3 However, these treatments are also associated with significant side effects that may limit their utility in some patients.4–6 The reported rates of neuropsychiatric adverse effects vary greatly, depending on the study design and methodology. In observational or prospective studies, depression has been reported in up to 44% of treated patients,7–9 although depression associated with PegIFN/RBV therapy was reported in only 20–30% of patients enrolled in registration trials.2, 3 When the incidence of depression, anxiety and anger is collectively analysed, the cumulative prevalence of neuropsychiatric symptoms can be as high as 57%.10
Mood disturbances induced by PegIFN/RBV may hinder adherence with the rigorous medical regimen, and can have deleterious effects on the patient’s personal, social, occupational and academic functioning.11–16 Moreover, IFN-induced depression and other psychiatric syndromes may be correlated with diminished rates of viral clearance.17, 18 Patients initially deferred from therapy because of comorbid psychiatric or substance use issues who were subsequently treated at a later date, have significantly higher rates of treatment discontinuation and had marginally lower rates of sustained virological response.19 A few studies have demonstrated that management with antidepressants, particularly SSRIs, administered pre-emptively or after the development of neuropsychiatric side effects, can improve outcomes.9, 20, 21 Thus, IFN-induced mood disturbance is disruptive for patients, may affect adherence, may result in diminished chances of viral clearance, but can be ameliorated with specific interventions. This underscores the importance of identifying risk factors for neuropsychiatric side effects and closely monitoring the development of these events throughout the course of treatment.
The most significant limitation of previous studies is the lack of generalizability to patients treated outside both clinical trials and VA medical centres. Many clinical trials reported low rates of psychiatric side effects (∼20%), but this is undoubtedly because of careful patient selection and stringent criteria that excluded individuals with mental health and substance use disturbance (MH/SUD). Other prospective clinical trials, which aimed to investigate potential mechanisms of IFN-induced mood disturbance, included mostly euthymic patients with no previous or current MH/SUD. For instance, studies excluded patients with past or present history of Axis I disorders (e.g. major depression, bipolar, dysthmia and anxiety) or who were taking psychiatric medication.22 Other studies excluded patients with past or current mild-to-severe MH/SUD, who were using psychotropic medications, were using any drugs or alcohol, or who had any previous suicide attempts.9, 23, 24 The advantage of excluding patients on these grounds is that there is less interference from pre-existing patient factors to allow for a more pristine, uncontaminated, association between IFN and mood dysfunction in euthymic patients. The disadvantage, however, is that the findings may not be applicable to patients with MH/SUD. Finally, a large portion of the data available on psychiatric symptoms associated with PegIFN/RBV is based on US veterans treated at VA facilities, and with the extremely high rates of MH/SUD in the veteran population,25, 26 these data are unlikely to be representative of individuals with HCV treated in university-based or tertiary care centres.
The aim of this study was to evaluate the prevalence of MH/SUD prior to PegIFN/RBV treatment, and the incidence and course of psychiatric symptoms during treatment, in a large cohort of patients treated under standard of care protocol in a university-based tertiary care setting.
This was a retrospective study of all consecutive patients with chronic hepatitis C who were treated with PegIFN/RBV, between February 2002 and December 2006 at the UNC Liver Program, a university-based tertiary care hepatology centre. Participants who were still on therapy or who were participating in a clinical trial during this time frame were excluded. Data were obtained during patient clinical visits and were entered in electronic medical records. For this study, we retrospectively collected data from all medical records available through the UNC healthcare system, such as hepatology, psychiatric, internal medicine and emergency room records. Data included: demographics, genotype, documented history of mental health or substance use, psychotropic medications at start of treatment, psychiatric symptoms during treatment (i.e. depression, anxiety and irritability) as defined below, onset of symptoms, and clinical management of psychiatric symptoms. Two research assistants (KS and AV) assisted the PI (DE) in reviewing medical records, and all reviewers adhered to standardized guidelines. Inter-rater reliability between two reviewers on 42 cases for six key variables ranged from κ = 0.90 to κ = 0.10 for a simple kappa coefficient. The Biomedical Institutional Review Board at the University of North Carolina at Chapel Hill approved this study.
The UNC hepatology clinic where new HCV patient referrals were seen was staffed with two attending hepatologists, two GI fellows and two doctor extenders. One of the attending hepatologists oversaw the decision-making process to determine patient eligibility for PegIFN/RBV therapy. Seventy-four percent of patients who were evaluated during the study period were seen by the two primary attending hepatologists, whereas the remaining 26% of patients were seen by one of four other hepatology attendings.
Medical histories, with full review of systems, were obtained during the initial interview. Because ongoing psychiatric illness and drug and alcohol abuse have been strong relative contraindications to treatment, clinicians specifically queried and recorded this information during pre-treatment clinical visits. Clinicians solicited information about the past and current drug and alcohol history, typically inquiring into type of substance, and frequency, duration and amount of substances used. Psychiatric hospitalizations, suicide attempts, suicidal ideation, psychotropic medication use, psychiatric care and diagnoses, are all queried to determine the patients eligibility for PegIFN/RBV treatment. Patients were typically deferred from treatment for the following reasons: (i) suicide attempt or psychiatric hospitalization in the past 5 years; (ii) ongoing active suicidal ideation; (iii) severe psychiatric disturbance that was not well- controlled or being managed by a mental health provider; (iv) ongoing significant lifestyle or psychosocial stressors and (v) alcohol or illicit drug abuse in the past 6 months. Aside from these exclusion criteria, patients treated under standard of care protocol may have had histories of suicide attempts or psychiatric hospitalizations prior to 5 years ago, past suicidal ideation, mild-to-moderate psychiatric problems, past or current use of psychotropic medications and methadone use.
The same nurse practitioner treated all 215 patients during the study period. Standard of care treatment was based on genotype; 48 weeks for genotype 1, and 24 weeks for genotypes 2 and 3. Patients were examined and monitored closely at weeks 2, 4, 6, 8 and then every 4 weeks thereafter. During all clinical visits, side effects were assessed and managed. Neuropsychiatric side effects, suicidal and homicidal ideation were monitored closely. The Center for Epidemiological Studies Depression Scale was administered and used for clinical assessment and management of depressive symptoms. The nurse practitioner was accessible via phone and page on evenings and weekends should problems arise. All of these visits, as well as unscheduled visits and records of patient phone calls, were documented in the electronic medical record and were available for review.
Definition of constructs
History of substance abuse was defined as documentation in the patient’s medical record, prior to initiating treatment, of a history of alcohol or drug abuse, as judged by the provider or reported by the patient. Drug abuse included all illicit substances and abuse of prescription medication. The following excerpts illustrate the type of documentation considered an endorsement of substance abuse: (i) ‘prior to 10/2003, he consumed 4–12 beers per day and more on weekends and days off’, (ii) ‘he has a history of heavy alcohol use in the past although decreased now for many years. He has a history of IV drug use in the past as well as intranasal cocaine use’.
History of depression was defined as any documentation in the patient’s medical record, prior to initiating treatment, of a history of unipolar or bipolar depressive symptoms, as judged by the provider or reported by the patient. Depressive symptoms ranged from mild to severe. History of suicide attempts or suicidal ideation was also captured under this variable. The following are two excerpts that illustrate the type of documentation considered an endorsement of a history of depression: (i) ‘patient does have underlying depression which is not adequately controlled’, (ii) ‘over the last several months, the patient has felt somewhat depressed and asked for an antidepressant medication and was started on Prozac with some improvement in his mood’.
History of anxiety was defined as any documentation in the patient’s medical record, prior to initiating treatment, of a history of anxiety symptoms, as judged by the provider or reported by the patient. Anxiety symptoms could be general, mild to severe, or include a specific anxiety diagnosis. The following two excerpts demonstrate the type of documentation considered an endorsement of a history of anxiety: (1) ‘he reports history of anxiety and depression disorder which is managed by his PCPs office. His anxiety is well controlled at current dose of Xanax’, (ii) ‘she does have a history of some mild anxiety and some mild depression, although my impression is that these are not contraindications to therapy in this situation’.
History of irritability was defined as any documentation in the patient’s medical record, prior to initiating treatment, of a history of irritability or anger problems, as judged by the provider or reported by the patient. The following two excerpts demonstrate the type of documentation considered an endorsement of a history of irritability: (i) ‘he admits to depression, difficulty managing anger outbursts and continues to consume alcohol on a regular basis’, (ii) ‘his major obstacles for obtaining treatment are his depression and mood issues. He was referred to Psychiatry and diagnosed with impulse control problems’.
On-treatment psychiatric symptoms
Depression on treatment was defined as any documentation in the patient’s medical record during treatment, where the patient self-reported, or the nurse practitioner indicated an increase in depressive symptoms or suicidal ideation. Symptoms ranged from mild to severe. The following two excerpts demonstrate the type of documentation considered an endorsement of depression on treatment: (i) ‘increased symptoms of depression and irritability including increased tearfulness, decrease in performing activities that she once enjoyed such as exercise’, (ii) ‘he reports mild increase in depression with tearfulness, which is unusual for him. He feels that he is more sensitive, he wakes up tearful; however, denies a sense of hopelessness or any suicidal ideation’.
Anxiety on treatment was defined as any documentation in the patient’s medical record during treatment, where the patient self-reported, or the nurse practitioner indicated an increase in symptoms of anxiety, nervousness or agitation. Anxiety ranged from mild to severe. The following two excerpts demonstrate the type of documentation considered an endorsement of anxiety on treatment: (i) ‘this patient was dose reduced secondary to increased anxiety, (ii) ‘experienced increasing shortness of breath over a period of 3–4 days that required evaluation at the local ER with overnight stay in the hospital… This incident was deemed to be a panic attack, probably IFN induced’.
Irritability on treatment was defined as any documentation in the patient’s medical record during treatment where the patient self-reported, or the nurse practitioner indicated an increase in symptoms of irritability, anger or hostility. Irritability ranged from mild to severe. The following two excerpts demonstrate the type of documentation considered an endorsement of irritability on treatment: (i) ‘he feels that he has slight increase in irritability. He has difficulty dealing with people and feels that he has alienated a few people’, (ii) ‘she reports her mood has slightly worsened particularly with increase in irritability. We will decrease her Pegasys to 135 μg daily secondary to increase in irritability and depression’.
Data were analysed in sas (Cary, NC, USA). Univariate statistics were used to describe the key characteristics of the study population (n =215). Mean, standard deviations, frequency counts and percentages were conducted for demographics, previous substance use and mental health issues, prior use of psychotropic medications, and the development and management of psychiatric side effects. Inter-rater reliability coefficients were calculated on three baseline and three treatment variables to ensure consistency among reviewers.
Kaplan–Meier survival analysis curves were generated to determine the proportion of patients who develop depression, anxiety and irritability on treatment, based on whether or not they had documented histories of these problems. The curves were compared with the log-rank test. P-values <0.05 were considered statistically significant.
Two hundred and fifteen (n =215) consecutive patients who met the above entry criteria were included in these analyses. Patients were 45.5 years old (range: 17–72), and 59% were male. Racial background was 79% Caucasian, 19% African-American and 2% from other races. Sixty-seven percent of patients were genotype 1, 31% were either genotype 2 or 3.
Prior to beginning treatment, 47% of patients (102 of 215) had a reported history of depression, 14% (31 of 215) reported a history of anxiety and 5% (10 of 215) reported a history of irritability or anger problems. Forty-eight percent (104 of 215) of patients reported a history of significant substance use/abuse. The total number of patients with any lifetime history of MH or SUD, prior to initiating therapy, was 67% (145 of 215).
Prior use of psychotropic medications
Thirty-nine percent (83 of 215) of patients reported taking a psychotropic medication at some point in the past or were still taking it at the time they were referred to the hepatology clinic. Of these 83 patients, 80% (66 of 83) continued on psychotropic medications throughout HCV therapy. The most common psychotropic medications patients were previously taking included: paroxetine (13%), bupropion (12%), sertraline (10%) and alprazolam (10%). Six of 83 patients had no information available regarding the psychotropic medications they were taking prior to therapy.
A small number (8%, 17 of 215) of patients were identified by the treating clinicians during a clinical visit as being at risk for developing IFN-induced mood disturbance, and these patients were subsequently prescribed a prophylactic antidepressant in preparation for IFN starting therapy. The most common medications prescribed at baseline were sertraline (47%), citalopram (29%) and fluoxetine (12%). Altogether, 37% (80 of 215) of patients were taking psychotropic medications at baseline prior to treatment.
Development of psychological symptoms on treatment
Depression. During pegIFN/RBV therapy, 46% (98 of 215) of patients endorsed some level of depressive symptoms. Of 102 patients with a history of depression, 63 (62%) developed depression on PegIFN/RBV therapy, but for the 113 patients with no history of depression, only 35 (31%) developed depression on treatment. Depression was first reported within the first 2 months of treatment (mean = 8 weeks, median = 6 weeks, range: 1–40, s.d. = 6.6). In patients with and without histories of depression, both groups show a dramatic increase in depression through week 12, with a gradual tapering off after that week (see Figure 1). Patients rarely developed depression after week 16. Figure 1 illustrates that patients with documented histories of depression are more likely to develop depression on PegIFN/RBV therapy, compared with patients with no history of depression.
Anxiety. Nineteen per cent (40 of 215) of patients endorsed symptoms consistent with anxiety during the course of treatment. Of 31 patients with a history of anxiety, 21 (68%) developed anxiety on PegIFN/RBV therapy. In 184 patients with no history of anxiety, only 19 (10%) developed anxiety on treatment. Like depression, anxiety developed within the first 2 months of treatment (mean = 9 weeks, median = 7 weeks, range: 2–48, s.d. = 8.7). Anxiety developed rapidly by week 2 in patients with histories of anxiety, and continued to develop for patients until week 12 (see Figure 2). The development of anxiety tapers off in both groups, such that only one patient developed anxiety after week 24. Figure 2 illustrates that patients with documented histories of anxiety are more likely to develop depression on PegIFN/RBV therapy compared with patients with no history of anxiety disturbance.
Irritability. Almost half of patients on treatment (46%; 99 of 215) reported symptoms of irritability and anger first developing within 2 months (mean = 9 weeks, median = 6, range: 1–40, s.d. = 7.6) of initiating treatment. Of 10 patients with a history of irritability, nine (90%) developed irritability on IFN therapy, while only 44% of patients with no history of irritability developed irritability on treatment. The inception and trajectory of irritability are similar for both groups, such that it developed primarily in the first 12 weeks of treatment, with very few patients developing irritability after week 20 (see Figure 3). Irritability developed rapidly by week 2 in both groups. Figure 3 illustrates that patients with documented histories of irritability are more likely to develop irritability on PegIFN/RBV therapy compared with patients with no history of irritability.
Management of psychological symptoms. Cumulatively, 64% (138 of 215) of patients indicated at least mild mood disturbance while undergoing PegIFN/RBV therapy that required an assessment by the nurse practitioner. When mood symptoms emerged, the nurse practitioner instituted one or more interventions (see Table 1). Of 138 patients who reported some mood disturbance, 60% (83 of 138) required an active intervention, while close monitoring was deemed sufficient for 40% (55 of 138) of patients. The most common intervention was the addition of a psychiatric medication (48%). This was prescribed by the nurse practitioner, the patient’s primary care doctor or a psychiatrist, if the patient was already receiving such care. Dose reductions occurred in 14%, and treatment had to be discontinued in 12% of patients. When dose reduction or discontinuation was recommended, it was often because the patient was experiencing other systemic side effects, in addition to mood disturbance.
Table 1. Interventions taken to manage mental health and substance use disturbance on interferon (IFN) therapy
Depression (n =98)
Anxiety (n =40)
Irritability (n =99)
Cumulative (n =138)
Current psychotropic medication increased
Psychotropic medication prescribed
Combination of increased and prescribed
IFN and/or ribavirin dose decreased
Discontinued from IFN therapy
This retrospective analysis of a large cohort of patients who were treated with pegylated interferon and ribavirin at a university-based tertiary care hepatology clinic represents an important observational study of the prevalence, course and management of psychiatric symptoms occurring during HCV therapy. This study expands upon previous research by characterizing pre-existing MH/SUD and the psychological problems encountered during PegIFN/RBV therapy in a cohort of patients who were treated during standard of care protocol in a non-VA treatment facility.
One criticism levelled against previous studies has been the stringent exclusion criteria for entry into research studies, and subsequently, that findings may not be applicable to the majority of patients with HCV, many of whom have MH/SUD. The expanded treatment population investigated in this study is likely to be more representative of the larger population of individuals with hepatitis C who will require treatment. In contrast to participants in many clinical trials, patients treated under standard of care protocol in this study had histories of mild-to-severe mood disorders, remote suicide attempts and psychiatric hospitalizations, histories of suicidal ideation, psychotropic medication use and methadone use. These patients, many of whom would not have met the rigorous standards required of clinical trials, had high rates of psychotropic medication use before and during treatment, and higher rates of psychiatric disturbance on anti-viral therapy, compared with previous descriptions in the literature.2, 3, 7–9 When compared with US veterans treated in VA facilities, the patients in this study appear to have lower rates of premorbid mental health and substance use disturbances, but approximately the same rate of psychiatric disturbance on treatment.27
Many of the patients referred for HCV anti-viral therapy have pre-existing or co-occurring MH/SUD. In a previous study on treatment deferral rates, we found that 75% of patients referred to our clinic were deferred from therapy,19 which is consistent with other reports in the literature.28–30 MH and SUD were the primary reasons for deferral, accounting for one-third of all deferrals. Thus, this study includes only those 25% of patients deemed to be appropriate treatment candidates, culled from this larger pool. Ironically, even in this ‘ideal’ treatment cohort without absolute contraindications to therapy, a large majority (67%) had lifetime prevalence rates of mild-to-severe MH/SUD. The high prevalence rate for MH/SUD in this study compared to others may be related to sample characteristics, inclusiveness of definitions of MH/SUD or our clinical setting. The disproportionate number of challenging patients treated in our clinic may reflect referral bias whereby patients with fewer co-morbidities receive therapy with community-based providers, and more difficult-to-treat patients were referred to our tertiary care clinic. Our definitions of histories of MH/SUD were relatively inclusive, capturing patients with mild-to-severe MH/SUD. Nevertheless, this cohort of patients with MH/SUD represents a large constituency of patients who will require anti-viral therapy in the future. Multidisciplinary models of care and interventions for patient selection and management will need to be further investigated.
The incidence of mood disturbance on treatment in this cohort was higher than most previously reported in the literature. Compared to incidence rates of 22–39% for depression in previous studies, 46% of patients in this study developed depressive symptoms. Irritability was just as common as depression, with almost half the patients endorsing irritability symptoms, a rate higher than previously reported. Patients with histories of depression were twice as likely to develop depression on treatment, and patients with histories of anxiety were six times as likely to develop anxiety as their non-anxious counterparts. Very few patients had documented pre-existing anger problems; however, almost all patients identified with irritable tendencies reported worsening of irritability or hostility on treatment. Irritability and hostility associated with PegIFN/RBV therapy have not received nearly the amount of attention that depression has; yet, these negative mood states can be just as disruptive as depression.31, 32 These findings underscore the importance of screening for a wide range of neuropsychiatric side effects on IFN therapy, including anxiety, irritability and hostility.
Determining when patients develop psychological side effects can assist clinical care. Depression, anxiety and irritability all emerged in this cohort primarily between weeks 2 and 12, irrespective of prior history. When patients develop mood disturbance, approximately 20% will develop it rapidly by week 4. Onset of mood disturbance was infrequent after 12 weeks, and by weeks 20–25, the risk of developing psychological symptoms dropped remarkably. Thus, patients should be most closely monitored for psychological symptoms as early as week 2, but once they pass the 20–25 weeks threshold, risk of developing these symptoms appears to diminish.
The management of psychological symptoms ranged from watchful observation for mild symptoms to dose reduction and/or treatment discontinuation for severe symptoms. The majority of patients (66%) who experienced mood disturbance needed specific interventions, usually the addition or adjustment of a psychotropic medication by our nurse practitioner, or the patient’s internist or psychiatrist. The medications most often prescribed were from the SSRI class of antidepressants. Furthermore, dose reductions and treatment discontinuations for MH/SUD problems occurred in 13% and 10% of patients respectively. Although these numbers are somewhat higher than other reports,10 they are not surprising given our sample characteristics.
Several studies have explored risk factors for the development of IFN-induced depression.8, 16, 33 When standardized diagnostic evaluations or depression screening instruments are utilized, studies suggest that current or recent depression, rather than lifetime depression, is associated with IFN-induced depression and that baseline depression may confer a higher risk for IFN-induced depression, even if that depression is minor or subclinical. Our retrospective analysis did not allow us to differentiate between lifetime depression and depression at the start of therapy. This distinction is relevant to the interpretation of our findings and comparison with previous studies. In all likelihood, many patients in our study could have suffered from mild or subclinical psychiatric syndromes at treatment initiation and therefore strengthened the association between ‘history of depression’ and the development of depression on treatment.
This study has limitations inherent in all retrospective analyses. The definitions of mental health and substance abuse disturbances were not based on structured interviews or screening instruments collected prospectively, but based on the clinical judgment of providers and patient self-reports. It is noteworthy to mention that while the definitions of what constituted a history of MH/SUD may have varied across providers pre-treatment, the same clinician treated all 215 patients, thus reducing variability in clinical judgment during treatment. The inter-rater reliability among the reviewers of the medical records was also excellent (r = 0.90–1.0). As diagnostic evaluations were not conducted, we were unable to define fully the spectrum of psychiatric disorders that were present in patients before and during PegIFN/RBV therapy, nor could we differentiate between a true IFN-induced mood disturbance and disturbance that may have resulted from an environmental stressor independent of the medication. This is actually difficult to differentiate not only retrospective studies, but in all studies. Despite these limitations, the findings of this study are of value to most providers who treat patients outside clinical trials and VA medical centres, and who make treatment decisions and manage psychiatric side effects based on clinical judgment, rather than by formal psychological evaluations.
There has been much trepidation to treat patients with previous or concurrent MH/SUD with PegIFN/RBV therapy. In this study of patients with a high prevalence rate of MH/SUD, many patients developed mild-to-severe mood disturbance, but most of these symptoms were manageable, and it was relatively rare that medications were reduced or discontinued for reasons of mood disturbance. Patients with pre-existing mood disturbance, on psychotropic medication, or receiving mental health care, may be more cognizant of their mood disturbance vulnerability, and perhaps the presence of this insight makes them easier to treat. They may be more adept at identifying early signs of negative mood, and many already have pharmacological and nonpharmacological strategies in place to cope with mood dysfunction. Finally, as many patients suffer adjustment disorders with depressed or anxious mood as a result of their HCV diagnosis, liver disease and perceived social stigma, the opportunity to receive potentially curative treatment engenders hope and optimism that may partially counteract the adverse side effects of anti-viral medication.
In conclusion, mental health and substance use disorders are highly prevalent in patients treated for hepatitis C in tertiary care centres outside research trials. Psychiatric symptoms associated with PegIFN/RBV, identified through clinical criteria, are common, develop early in the treatment course, and are usually manageable without the need for treatment discontinuation in most patients.
Declaration of personal interests: None. Declaration of funding interests: This study was funded in part by an NIH K24 mentoring award DK066144 (MWF).