SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References

The currently recommended treatment for hepatitis C virus (HCV) infection is pegylated interferon alfa (PEG-INF) and ribavirin, which can be difficult to tolerate. More information about predicting sustained virologic response (SVR) may allow more informed treatment decisions to be made. This retrospective observational cohort study identified predictors of SVR to PEG-INF and ribavirin in routine medical practice at 121 Department of Veterans Affairs facilities. Among 5,944 patients infected with HCV genotypes 1, 2, or 3 who had been treated with PEG-INF and ribavirin, SVR rates were 20%, 52%, and 43%, respectively, and discontinuation rates were 68% (prior to 48 weeks), 34% (24 weeks), and 41% (24 weeks), respectively. In multivariate analysis, significant predictors of decreased likelihood of genotype 1 patients having an SVR were being African American, clinical liver disease, diabetes, low cholesterol, low hemoglobin, low platelet count, and treatment at a low-volume facility. Predictors of increased likelihood of genotype 1 patients having an SVR were low-level HCV viremia, elevated ALT quotient, and receiving PEG-INF 2A (rather than 2B). For genotype 2 patients, increasing body mass index, prior use of interferon, and low platelet count were negative predictors; only low-level HCV viremia was a positive predictor. For genotype 3 patients, only receiving PEG-INF 2A affected the likelihood of an SVR; its effect was positive. Conclusion: Among patients for whom HCV treatment is initiated during routine medical care, multiple factors including form of PEG-INF received affect the SVR rate for genotype 1 patients. Few of these factors affect the rate for genotype 2 patients, and even fewer do so for genotype 3 patients. (HEPATOLOGY 2007.)

In the United States, the recent National Health and Nutrition Examination Survey estimated that approximately 3.2 million people have chronic hepatitis C virus (HCV) infection.1 In the United States, HCV infection accounts for approximately 40% of all those with chronic liver disease, is the most frequent indication for liver transplantation, and causes 8,000-10,000 deaths annually.2 Moreover, morbidity and mortality from HCV infection are expected to increase dramatically.3

Based on 3 pivotal clinical trials,4–6 the current standard of care for HCV treatment is a combination of pegylated interferon alfa (PEG-INF) and ribavirin.7 Although this treatment promises to reduce HCV morbidity and mortality,8 it can be difficult to tolerate. About 60% of patients develop fatigue and flu-like symptoms, 20%-30% experience depression, 10%-20% develop neutropenia, and 10%-20% develop anemia.9–11 The rate of discontinuing treatment in clinical trials because of side effects has ranged from 14% to 21%4–6 and is likely higher in routine medical practice. Given the likelihood of side effects and the possibility of discontinuation, additional information about predicting sustained virologic response (SVR) may allow clinicians and patients to make more informed treatment decisions.

The Department of Veterans Affairs (VA) provides comprehensive medical care to a large HCV-infected population, which makes it possible to assess SVR rates and predictors in routine medical practice.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References

We used the VA Clinical Case Registry: Hepatitis C (CCR:HepC), an extract of VA electronic medical record information for HCV-infected patients receiving care at any VA medical facility. The CCR:HepC contains data on laboratory results, pharmacy information, height/weight, and International Classification of Diseases, Ninth Revision (ICD9) diagnosis codes from inpatient stays, outpatient visits, and problem lists from at least January 1, 1996. The study protocol was approved by the Stanford University Institutional Review Board and the VA Palo Alto Health Care System Research and Development Committee.

A patient was included in the cohort if he or she: (1) had started VA-prescribed PEG-INF with ribavirin by September 30, 2003; (2) had a VA-detected quantitative HCV RNA result less than 3 years before starting PEG-INF; and (3) had been infected with HCV genotypes 1, 2, or 3. The date of a patient's first VA PEG-INF prescription was defined as his or her HCV treatment start date. For patients who started treatment on September 30, 2003, laboratory data extended at least 34 weeks beyond a 48-week course of treatment. Gaps in treatment were identified as the difference between the last date covered by previous prescriptions and the fill date for the next prescription. Patients who had gaps greater than 45 days were considered to have received sequential treatment courses; we consider only the first course. Exclusion criteria were coinfection with human immunodeficiency virus (HIV), liver transplantation prior to start date, treatment with PEG-INF in a VA clinical trial, and a switch in the form of PEG-INF treatment.

Patients were considered responders if they had an SVR, defined as HCV RNA being undetectable in all follow-up HCV RNA tests after the treatment end date, including at least 1 test more than 12 weeks after that date. The treatment end date was defined as the later of the last date covered by the final PEG-INF prescription or of all PEG-INF prescriptions. We allowed a test 12 weeks after the treatment end date, rather than the usual 24 weeks, because an analysis of 624 patients with end-of-treatment responses showed that 98% of relapses occur within 12 weeks of cessation of therapy.12 Patients with undetectable HCV RNA after the treatment end date but who were not tested at least 12 weeks after that date were excluded from the cohort. Patients not tested for HCV RNA after the treatment end date or with detectable HCV RNA at any time after that date were considered nonresponders.

Baseline weight, height, and laboratory test results (except hepatitis B and cholesterol) were defined as the VA results within 1 year of starting treatment that were closest to the treatment start date. Patients were considered infected with hepatitis B virus if they had tested positive for either hepatitis B surface antigen or hepatitis B e antigen at any time prior to the start date. We determined whether a patient's most recent cholesterol result within 3 years prior to the start date was less than 130 mg/dl and whether a patient had a baseline alanine aminotransferase (ALT) greater than 3 times the upper limit of normal (ALT quotient > 3). We also determined whether each patient had a baseline relative laboratory value that was a contraindication for HCV treatment, defined in the 2003 VA treatment recommendations: hemoglobin < 13 g/dl for men and < 12 g/dl for women, absolute neutrophil count ≤ 1.5 k/μl, platelet count ≤ 75 k/μl, and creatinine > 1.5 mg/dl.13 We considered patients whose HCV RNA test results at their facility was above the quantification limit (e.g., >850,000 IU/ml) as having that upper limit (850,000). Patients with HCV RNA of less than 500,000 IU/ml were considered to have low-level HCV viremia.

Several additional measures captured conditions that may affect treatment response.14, 15 Using all VA ICD9 codes available prior to the start date, we categorized a patient as having a history of cirrhosis, decompensated liver disease (esophageal varices, hepatic encephalopathy, hepatorenal syndrome, or portal hypertension), or depression. A patient was categorized as having diabetes based on an ICD9 code or a baseline glucose greater than 200 mg/dl. We categorized a patient as having “recent” alcohol abuse, hard drug use (opiates, cocaine, or amphetamines), or socioeconomic instability (homelessness, poverty, or unemployment) based on ICD9 codes within 1 year prior to the start date.

Patients with a VA prescription for nonpegylated interferon prior to the start date were considered HCV treatment experienced.

Additional patient demographic variables measured were age at start date, sex, and race. The CCR:HepC did not include information on the race of approximately 30% of the cohort patients.

Treatment Initiation and Course.

We determined each patient's starting and lowest doses of PEG-INF and ribavirin. PEG-INF 2A doses were categorized as 180 μg/week or less. To allow for clinician rounding of patient weight and dose, PEG-INF 2B doses were categorized as <0.9, 0.9-1.1, 1.2-1.3, or ≥1.4 μg/(kg · wk). Ribavirin was categorized as at or above recommended levels according to genotype as follows: genotype 1, 1,000 mg/day for patients ≤ 75 kg and 1,200 mg/day for patients > 75 kg; genotypes 2 and 3, 800 mg/day.11, 13

Treatment duration was determined by calculating the cumulative days of supply of PEG-INF prescriptions and categorized as <60%, 60%-79%, 80%-99%, or ≥100% of the recommended duration of treatment (48 weeks for genotype 1, 24 weeks for genotypes 2 and 3).9, 10, 13 The medication possession ratio (MPR) was calculated by dividing the cumulative days of PEG-INF supply by the total number of days of PEG-INF supply needed to have continuous treatment from the start date to the end date (0.5 indicates sufficient medication for half the prescribed doses).

The prescribing clinic of each patient's first PEG-INF prescription was categorized as gastroenterology/hepatology, infectious disease, or internal medicine/other. The prescribing facility was further categorized as low (<25), medium (25-99), or high (≥100) volume according to how many of the cohort patients received their first prescription there. Finally, we determined whether a patient received a hematopoietic growth factor (granulocyte colony-stimulating factor, granulocyte macrophage colony-stimulating factor, epoetin alfa, or darbepoetin alfa) while on treatment.

Statistical Analysis.

The Pearson chi-square test was used for univariate comparisons of responder and nonresponder characteristics. Predictors with a P value < .1 in univariate analyses for either the entire cohort or a genotype subgroup were included in multivariate models to predict SVR. Models were estimated for the entire cohort and separately by genotype. One set of models included baseline variables and treatment initiation variables, those likely to be available to clinicians and patients when deciding whether to initiate treatment. Another set included treatment course variables. We used a backwards stepwise approach to find the most important SVR predictors. At each step, the candidate predictor with the largest P value was dropped until all remaining variables had P values ≤ .05. Data were analyzed using SAS version 8.2 (SAS Institute, Cary NC).

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References

We identified 9,022 patients who started PEG-INF/ribavirin treatment. Patients were excluded from the cohort if their HCV genotype was not electronically available (452), their genotype was other than 1, 2, or 3 (49), their baseline HCV RNA was undetectable (98), the results of their baseline HCV RNA test was not electronically available (1,698), their baseline HCV RNA test had no electronic results (345), they had HIV infection (203), they had had a liver transplant (26), they were enrolled in a PEG-INF/ribavirin clinical trial (59), they had switched PEG-INF form (71), or they had undetectable HCV RNA only on tests less than 12 weeks after the treatment end date (77).

Baseline, treatment initiation, and treatment course characteristics of the final cohort of 5,944 patients appear in Table 1. More than 95% of the identified African American patients had genotype 1. Thirteen percent of patients had at least 1 laboratory result that was a relative contraindication to PEG-INF/ribavirin treatment.

Table 1. SVR Rates for Baseline and Treatment Characteristics of Cohort Starting PEG-INF/Ribavirin
 Cohort (N = 5944) Cohort (N = 5944)
  1. Abbreviations: ALT, alanine aminotransferase; ANC, absolute neutrophil count; BMI, body mass index; GCSF, granulocyte colony-stimulating factor; GMCSF, granulocyte macrophage colony-stimulating factor; GT, genotype; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; IQR, interquartile range; MPR, medication possession ratio; SD, standard deviation.

Baseline Characteristic Facility volume 
Sex, male96% Low9%
Age (years), mean (SD)51.0 (5.7) Medium56%
Race  High35%
 African American17%Initiating clinic 
 White49% Gastroenterology/hepatology78%
 Other2% Infectious disease13%
 Not recorded31% Internal medicine/other9%
Weight (kg), mean (SD)91.7 (17.8)Treatment course 
BMI (kg/m2), mean (SD)29.1 (5.1)Lowest PEG-INF 2A dose 
Clinical liver disease  <180 μg/week6%
 No cirrhosis/no decompensation83% 180 μg/week94%
 Cirrhosis/no decompensation14%Lowest PEG-INF 2B dose (μg/kg/week) 
 Decompensated liver disease4% <0.95%
Depression49% 0.9–1.112%
Diabetes21% 1.2–1.322%
Recent lifestyle hazards32% ≥1.461%
Prior nonpegylated interferon12%Percentage of recommended PEG-INF duration 
HCV genotype   
 180% <60% (<29 weeks GT1; <15 weeks GT2/3) 
 212% 43%
 38% 60%–79% (29–38 weeks GT1; 15–19 weeks GT2/3) 
HCV RNA above local limit of quantification27% 8%
HCV RNA (IU/mL), median (IQR)763,000 (421,000–1,510,000) 80%–99% (39–47 weeks GT1; 20–23 weeks GT2/3) 
HBsAg or HBeAg1% 12%
ALT quotient, mean (SD)1.8 (1.5) ≥100% (≥48 weeks GT1; ≥24 weeks GT2/3) 
Cholesterol (mg/dl), mean (SD)172.2 (36.9) 38%
Creatinine (mg/dl), mean (SD)1.0 (.4)PEG-INF MPR 
Hemoglobin (g/dl), mean (SD)15.1 (1.4) <60%0%
Platelets (k/μ1), mean (SD)201.8 (69.1) 60%–79%3%
ANC (k/μ1), mean (SD)3.8 (1.7) 80%–89%8%
Treatment initiation  ≥90%88%
Form of PEG-INF Ribavirin dosing 
 2A35% All doses at/above recommended62%
 2B65% Starting dose below recommended29%
Starting PEG-INF 2A dose  Dose reduced below recommended9%
 <180 μg/week2%Ribavirin MPR 
 180 μg/week98% <60%4%
Starting PEG-INF 2B dose (μg/kg/week)  60%–79%11%
 <0.91% 80%–89%13%
 0.9–1.19% ≥90%72%
 1.2–1.320%Received epoetin/darbepoetin15%
 ≥1.470%Received GCSF/GMCSF7%
Starting ribavirin dose   
 Less than recommended29%  
 At/above recommended71%  

One third of patients were treated with PEG-INF 2A, 98% of whom started at the recommended dose of 180μg/week. Seventy percent of patients treated with PEG-INF 2B started on at least 1.4 μg/(kg · wk), and an additional 20% started on at least 1.2 μg/(kg · wk). Patients on PEG-INF 2A were more likely to receive treatment for the recommended duration than were patients on PEG-INF 2B (41% vs. 36%, P < .001).

One third of genotype 1 patients did not start on the recommended ribavirin dose, and only half started and remained on that dose. Almost all the patients who had genotypes 2 and 3 (97%) started and remained on the recommended ribavirin dose. While on PEG-INF, more than 85% of patients had a ribavirin MPR of at least 80%.

Only 32% of patients who had genotype 1 received the recommended 48-week course. Nearly two-thirds of patients who had genotypes 2 or 3 (66% and 59%, respectively) received at least the recommended 24-week course. Most patients had a high MPR for PEG-INF (≥80%).

Univariate Analyses of SVR Rates.

Overall, 1,551 patients had an SVR, for a rate of 26 % (Table 2); 20% of patients with genotype 1, 52% of patients with genotype 2, and 43% of patients with genotype 3 had an SVR (P < .001).

Table 2. SVR Rates for Baseline and Treatment Characteristics of Cohort Starting PEG-INF/Ribavirin for Entire Cohort and by Genotype
 Entire CohortGenotype 1Genotype 2Genotype 3
SVR RateP ValueSVR RateP ValueSVR RateP ValueSVR RateP Value
  1. Abbreviations: ALT, alanine aminotransferase; ANC, absolute neutrophil count; GCSF, granulocyte colony-stimulating factor; GMCSF, granulocyte macrophage colony-stimulating factor; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; IQR, interquartile range; MPR, medication possession ratio; SD, standard deviation.

Overall26% (1,551/5,944) 20% (973/4,755) 52% (376/720) 43% (202/469) 
Baseline Characteristics        
Sex .33 .14 .66 .51
 Female29% (68/236) 25% (46/186) 48% (16/33) 35% (6/17) 
 Male26% (1,483/5708) 20% (927/4,569) 52% (360/687) 43% (196/452) 
Age (years) .07 .39 .05 .65
 <4033% (48/145) 24% (28/116) 79% (15/19) 50% (5/10) 
 40–6926% (1,497/5,764) 20% (941/4,609) 52% (359/696) 43% (197/459) 
 ≥7017% (6/35) 13% (4/30) 40% (2/5) — (0/0) 
Race <.001 <.001 .16 .95
 African American15% (159/1,038) 14% (140/991) 38% (13/34) 46% (6/13) 
 White30% (834/2,891) 23% (497/2,188) 52% (210/405) 43% (127/298) 
 Other22% (29/148) 15% (18/119) 39% (7/18) 36% (4/11) 
 Not recorded28% (529/1,867) 22% (318/1,457) 56% (146/263) 44% (65/147) 
Weight .27 .97 .03 .84
 ≤75 kg28% (260/938) 21% (153/740) 62% (69/112) 44% (38/86) 
 >75 kg26% (1,284/4,942) 21% (816/3,959) 50% (305/604) 43% (163/379) 
Body Mass Index .25 .62 .01 .77
 BMI <25 kg/m228% (336/1,196) 22% (208/955) 63% (82/131) 42% (46/110) 
 BMI 25–29 kg/m226% (646/2,472) 20% (402/1,985) 53% (161/304) 45% (83/183) 
 BMI ≥30 kg/m225% (555/2,178) 21% (356/1,734) 47% (128/275) 42% (71/169) 
Clinical liver disease <.001 <.001 .09 .30
 No cirrhosis/no decompensation28% (1,355/4,912) 22% (857/3,925) 54% (336/625) 45% (162/362) 
 Cirrhosis/no decompensation20% (164/817) 15% (100/657) 41% (31/76) 39% (33/84) 
 Decompensated liver disease15% (32/215) 9% (16/173) 47% (9/19) 30% (7/23) 
Depression .24 .52 .06 .76
 Yes25% (742/2,920) 21% (506/2,429) 49% (178/365) 42% (97/229) 
 No27% (809/3,024) 20% (437/2,326) 56% (198/355) 44% (105/240) 
Diabetes <.001 <.001 .15 .19
 Yes20% (242/1,221) 16% (164/1,035) 46% (50/109) 36% (28/77) 
 No29% (1,309/4723) 22% (809/3,720) 53% (326/611) 44% (174/392) 
Recent lifestyle hazards .01 .04 .26 .07
 Yes24% (454/1,894) 19% (282/1,505) 49% (111/226) 37% (61/163) 
 No27% (1,097/4,050) 21% (691/3,250) 54% (265/494) 46% (141/306) 
Prior nonpegylated interferon <.001 .006 <.001 .05
 Yes18% (128/703) 16% (95/588) 27% (16/60) 31% (17/55) 
 No27% (1,423/5,241) 21% (878/4,167) 55% (360/660) 45% (185/414) 
HCV RNA above local limit of quantification .01 <.001 .28 .98
 Yes24% (386/1,622) 17% (212/1,259) 50% (133/268) 43% (41/95) 
 No27% (1,165/4,322) 22% (761/3,496) 54% (243/452) 43% (161/374) 
HCV RNA <.001 <.001 .003 .45
 <500,000 IU/ml32% (562/1,744) 27% (367/1,378) 62% (109/176) 45% (86/190) 
 ≥500,000 IU/ml24% (986/4,189) 18% (604/3,370) 49% (266/541) 42% (116/278) 
HBsAg or HBeAg .76 .30 .08 .28
 Yes25% (18/73) 26% (16/62) 17% (1/6) 20% (1/5) 
 No26% (1,413/5,384) 20% (879/4,294) 52% (344/657) 44% (190/433) 
ALT quotient <.001 <.001 .06 .82
 ≤325% (1,263/5,108) 19% (804/4,144) 50% (292/579) 43% (167/385) 
 >336% (273/760) 29% (162/550) 60% (77/129) 42% (34/81) 
Cholesterol .002 <.001 .42 .40
 <130 mg/dl21% (118/561) 14% (55/405) 46% (17/37) 39% (46/119) 
 ≥130 mg/dl28% (1,285/4,746) 22% (834/3,838) 53% (324/614) 43% (127/294) 
Creatinine (mg/dl) .93 .71 .38 .99
 ≤1.526% (1,433/5,461) 21% (898/4,357) 52% (347/666) 43% (188/438) 
 >1.527% (20/75) 19% (11/59) 67% (6/9) 43% (3/7) 
Hemoglobin (g/dl) <.001 .001 .12 .33
 <13 male/<12 female18% (76/419) 14% (48/347) 42% (20/48) 33% (8/24) 
 ≥13 male/≥12 female27% (1,454/5,432) 21% (912/4,333) 53% (352/662) 43% (190/437) 
Platelets (k/μ1) <.001 <.001 .02 .05
 ≤7511% (16/146) 7% (8/111) 26% (5/19) 19% (3/16) 
 >7527% (1,513/5,703) 21% (951/4,566) 53% (367/691) 44% (195/446) 
ANC (k/μ1) <.001 .008 .32 .13
 ≤1.516% (45/273) 14% (35/253) 38% (5/13) 71% (5/7) 
 >1.527% (1,400/5,277) 21% (871/4,185) 52% (347/662) 42% (182/430) 
Treatment initiation        
Form of PEG-INF <.001 <.001 .75 .002
 PEG-INF 2A31% (644/2,091) 25% (402/1,632) 53% (152/287) 52% (90/172) 
 PEG-INF 2B24% (907/3,853) 18% (571/3,123) 52% (224/433) 38% (112/297) 
Starting PEG-INF 2A dose (μg/week) .003 .002 .14 .62
 <1808% (3/37) 0% (0/29) 20% (1/5) 66% (2/3) 
 18031% (641/2,054) 25% (402/1,603) 54% (151/282) 52% (88/169) 
Starting PEG-INF 2B dose (μg/kg/week) .48 .43 .14 .42
 <0.918% (8/44) 10% (3/29) 45% (5/11) 0% (0/4) 
 0.9–1.121% (74/352) 16% (44/275) 36% (14/39) 42% (16/38) 
 1.2–1.324% (186/760) 18% (111/612) 58% (52/90) 40% (23/58) 
 ≥1.424% (634/2,647) 19% (410/2,163) 52% (151/290) 38% (73/194) 
Starting ribavirin dose <.001 .004 .84 .22
 Less than recommended19% (320/1,725) 18% (315/1,714) 56% (5/9) 0% (0/2) 
 At/above recommended30% (1,227/4,163) 22% (654/2,985) 52% (371/711) 43% (202/467) 
Facility volume .01 .02 .39 .44
 Low21% (117/549) 17% (76/448) 44% (25/57) 36% (16/44) 
 Medium27% (899/3315) 22% (585/2,683) 52% (202/386) 46% (112/246) 
 High26% (535/2,080) 19% (312/1,624) 54% (149/277) 41% (74/179) 
Initiating clinic .86 .46 .18 .76
 Gastroenterology/hepatology27% (987/3,677) 22% (626/2,904) 50% (236/472) 42% (125/301) 
 Infectious disease26% (158/610) 19% (95/497) 62% (42/68) 47% (21/45) 
 Internal Medicine/other27% (123/451) 21% (74/354) 54% (31/57) 45% (18/40) 
Treatment course        
Lowest PEG-INF 2A dose (μg/week) .73 .98 .38 .62
 <18032% (39/121) 25% (24/97) 64% (9/14) 60% (6/10) 
 18031% (605/1,970) 25% (378/1,535) 52% (143/273) 52% (84/162) 
Lowest PEG-INF 2B dose (μg/week) .07 .33 .02 .97
<0.918% (34/184) 15% (21/144) 29% (7/24) 38% (6/16) 
 0.9–1.123% (106/466) 19% (69/371) 42% (21/50) 36% (16/45) 
 1.2–1.327% (225/848) 20% (140/686) 61% (59/97) 40% (26/65) 
 ≥1.423% (537/2,305) 18% (338/1,878) 52% (135/259) 38% (64/168) 
PEG-INF duration (percent of recommended) <.001 <.001 <.001 <.001
 <60%4% (93/2,556) 3% (69/2273) 10% (17/163) 6% (7/120) 
 60–79%15% (68/451) 11% (42/394) 42% (13/31) 50% (13/26) 
 80–99%37% (254/692) 35% (207/590) 50% (27/54) 42% (20/48) 
 ≥100%51% (1,136/2,245) 44% (655/1,498) 68% (319/472) 59% (162/275) 
PEG-INF MPR <.001 <.001 .33 .36
 <60%15% (4/27) 14% (3/22) 33% (1/3) 0% (0/2) 
 60–79%12% (23/198) 5% (8/160) 35% (7/20) 44% (8/18) 
 80–89%27% (134/498) 20% (80/400) 58% (30/52) 52% (24/46) 
 ≥90%27% (1,390/5,221) 21% (882/4,173) 52% (338/645) 42% (170/403) 
Ribavirin dosing <.001 <.001 .98 .07
 All at/above recommended29% (1,070/3,637) 20% (507/2,482) 52% (362/694) 44% (201/458) 
 Starting below recommended19% (320/1,725) 18% (315/1,714) 56% (5/9) 0% (0/2) 
 Reduced below recommended30% (157/529) 29% (147/503) 53% (9/17) 11% (1/9) 
Ribavirin MPR <.001 .05 .20 .03
 <60%15% (39/255) 13% (28/214) 32% (6/19) 28% (5/18) 
 60–79%27% (167/630) 20% (103/503) 53% (41/77) 46% (23/50) 
 80–89%29% (219/755) 22% (133/609) 59% (48/82) 58% (38/65) 
 ≥90%26% (1,126/4,304) 21% (709/3,429) 52% (281/542) 40% (136/336) 
Received epoetin/darbepoetin <.001 <.001 .009 .08
 Yes36% (328/915) 31% (248/788) 65% (58/89) 56% (22/39) 
 No25% (1,223/5,029) 18% (725/3,967) 50% (318/631) 42% (180/430) 
Received GCSF/GMCSF .003 .005 .17 .03
 Yes33% (139/433) 26% (95/363) 63% (24/38) 61% (20/33) 
 No27% (1,412/5,511) 20% (878/4,392) 52% (352/682) 42% (182/436) 

SVR rate differed according to the baseline variables of race, recent alcohol abuse, recent hard drug use, and recent socioeconomic instability. Because the latter 3 measures were collinear, we created a composite variable indicating any of these lifestyle hazards. Fewer patients with diabetes, clinical liver disease, prior use of nonpegylated interferon, or low cholesterol at baseline had an SVR. The percentage of patients with a hematologic relative laboratory contraindication at baseline who had an SVR was smaller than that of patients without a contraindication. A higher percentage of patients with genotypes 2 or 3, an elevated ALT quotient, or low-level HCV viremia at baseline had an SVR. The rate of SVR did not differ according to the other baseline characteristics.

SVR rate differed by many treatment characteristics. The rate was higher for patients receiving PEG-INF 2A than for patients receiving PEG-INF 2B. It differed depending on whether the starting doses of PEG-INF 2A and ribavirin were at recommended levels, but not the starting dose of PEG-INF 2B. There was a clear positive relationship between duration of PEG-INF treatment and SVR rate. The SVR rate was lower among patients with MPRs of less than 80% for PEG-INF and less than 60% for ribavirin. Receiving hematopoietic growth factors during treatment was associated with an increased SVR rate. The percentage of patients treated at low-volume facilities who had an SVR was smaller than that of patients treated at higher-volume facilities (medium- and high-volume facilities were combined).

Multivariate Analyses.

In the backwards stepwise multivariate model for the entire cohort using the baseline and initial treatment characteristics, predictors of decreased likelihood of an SVR included African American race, clinical liver disease, diabetes, recent lifestyle hazards, prior use of nonpegylated interferon, low cholesterol, low hemoglobin, low platelet count, starting on a PEG-INF 2A dose of less than 180 μg/week, and treatment at a low-volume facility (Table 3). Predictors of increased likelihood of an SVR included genotypes 2 or 3, low-level HCV viremia, elevated ALT quotient, and treatment with PEG-INF 2A rather than PEG-INF 2B.

Table 3. Significant SVR Predictors in Backwards Deletion Regression Models
 Baseline and Treatment Initiation VariablesBaseline, Treatment Initiation and Treatment Course Variables
Odds Ratio (95% CI)P ValueOdds Ratio (95% CI)P Value
  1. Abbreviations: ALT, alanine aminotransferase; CI, confidence interval; Hgb, hemoglobin; MPR, medication possession ratio.

Baseline characteristics    
Race    
 African American0.57 (0.47–0.70)<.0010.62 (0.49–0.78)<.001
 White1.001.00
 Other0.57 (0.36–0.91).020.47 (0.28–0.77).003
 Not recorded1.02 (0.88–1.18).821.02 (0.86–1.20).84
Clinical liver disease    
 No cirrhosis/no decompensation1.001.00
 Cirrhosis/no decompensation0.70 (0.57–0.86)<.0010.66 (0.53–0.84)<.001
 Decompensated liver disease0.55 (0.36–0.84).0060.56 (0.35–0.90).02
Diabetes0.76 (0.64–0.91).0020.76 (0.63–0.92).005
Lifestyle hazards0.85 (0.74–0.98).03
Prior nonpegylated interferon0.66 (0.53–0.83)<.0010.67 (0.52–0.87).002
HCV genotype    
 11.001.00
 23.80 (3.16–4.55)<.0012.67 (2.17–3.30)<.001
 32.53 (2.02–3.17)<.0011.77 (1.37–2.28)<.001
HCV RNA <500,000 IU/ml1.77 (1.54–2.04)<.0012.01 (1.71–2.36)<.001
ALT quotient >3 × ULN1.56 (1.29–1.87)<.0011.57 (1.27–1.94)<.001
Cholesterol <130 mg/dl0.73 (0.58–0.92).0090.71 (0.55–0.93).01
Hgb <13 g/dl male/<12 g/dl female0.73 (0.55–0.97).03  
Platelets ≤ 75 k/μ10.42 (0.24–0.73).002  
Treatment initiation    
PEG-INF form    
 2A1.51 (1.32–1.73)<.0011.40 (1.20–1.64)<.001
 2B1.001.00
Starting PEG-INF 2A dose <180 μg/week0.17 (0.04–0.75).020.12 (0.03–0.57).006
Low-volume facility0.71 (0.56–0.90).0050.69 (0.53–0.90).006
Treatment course    
PEG-INF duration (percent of recommended)    
 <60%  0.04 (0.03–0.05)<.001
 60–79%  0.20 (0.15–0.27)<.001
 80–99%  0.63 (0.52–0.77)<.001
 ≥100%  1.00 
Ribavirin MPR <60%  0.59 (0.39–0.90).01
Received epoetin/darbepoetin  1.27 (1.05–1.52).01

When we expanded the backwards stepwise model to allow treatment course variables, most of the variables that had entered the baseline and treatment initiation model also entered the expanded model; the exceptions were lifestyle hazards, low hemoglobin, and low platelet count. Among the treatment course characteristics, receiving epoetin/darbepoetin increased the likelihood of an SVR, whereas a ribavirin MPR of less than 60% reduced this likelihood. The chance of an SVR increased with an increase in PEG-INF duration (across the 4 categories of PEG-INF duration). However, we cannot have confidence in these estimates, as interim results on HCV RNA response might influence treatment duration, which would cause duration variables to correlate with treatment response, leading to biased estimates.

Results of the genotype-specific models differed by genotype. When only baseline and treatment initiation variables were allowed, SVR predictors for patients with genotype 1 were comparable to those for the entire cohort except that recent lifestyle hazards, prior nonpegylated interferon, and starting PEG-INF dose of less than 180 μg/week no longer met the inclusion criteria (Table 4). In contrast, for genotype 2 patients, only low-level HCV viremia was a positive predictor, and increasing body mass index (BMI), prior nonpegylated interferon and low platelet count were negative predictors. For genotype 3 patients, only PEG-INF form entered the model; treatment with PEG-INF 2A increased the likelihood of an SVR relative to PEG-INF 2B. The persistence of PEG-INF form as a predictor in the multivariate models for genotypes 1 and 3 is consistent with univariate comparisons (Table 2). Patients with genotype 1 or genotype 3 treated with PEG-INF 2A were significantly more likely to achieve an SVR than were patients who received PEG-INF 2B. For genotype 2 patients, SVR rate did not differ by PEG-INF form.

Table 4. Significant SVR Predictors by Genotype in Backwards Deletion Regression Models Including Baseline and Treatment Initiation Characteristics
 Genotype 1 (n = 4,141)Genotype 2 (n = 697)Genotype 3 (n = 469)
Odds Ratio (95% CI)P ValueOdds Ratio (95% CI)P ValueOdds Ratio (95% CI)P Value
  1. Abbreviations: ALT, alanine aminotransferase; CI, confidence interval; Hgb, hemoglobin.

Baseline characteristics      
Race      
 African American0.55 (0.44–0.69)<.001    
 White1.00    
 Other0.56 (0.33–0.98).04    
 Not recorded0.95 (0.80–1.13).57    
Body mass index (kg/m2)      
 <25  1.00   
 25–29  0.67 (0.44–1.04).07  
 ≥30  0.55 (0.36–0.85).008  
Clinical liver disease      
 No cirrhosis/no decompensation1.00    
 Cirrhosis/no decompensation0.64 (0.50–0.82)<.001    
 Decompensated liver disease0.41 (0.24–0.70).001    
Diabetes0.74 (0.61–0.91).003    
Prior nonpegylated interferon  0.31 (0.17–0.56)<.001  
HCV RNA <500,000 IU/ml1.88 (1.60–2.22)<.0011.80 (1.26–2.57).003  
ALT quotient >31.65 (1.32–2.05)<.001    
Cholesterol <130 mg/dl0.60 (0.44–0.82).001    
Hgb <13 g/dl male/<12 g/dl female0.65 (0.46–0.92).01    
Platelets ≤75 k/μ10.35 (0.17–0.75).0070.29 (0.11–0.79).02  
Treatment initiation      
PEG-INF form      
 2A1.55 (1.32–1.81)<.001  1.81 (1.24–2.65).002
 2B1.00  1.00
Low-volume facility0.71 (0.54–0.94).02    

PEG-INF 2B was FDA approved and available in the VA system before PEG-INF 2A was. To investigate the possibility that PEG-INF form was a proxy for time and clinician experience, we limited the analysis to patients starting treatment in 2003, when both forms were available. In univariate analysis, patients receiving PEG-INF 2A remained more likely to achieve an SVR than did patients receiving PEG-INF 2B (31% [565 of 1,819] vs. 25% [308 of 1,222], P ≤ .001). Genotype 1 patients starting treatment in 2003 with PEG-INF 2A also remained more likely to achieve an SVR than those receiving PEG-INF 2B (25% [350 of 1,407] vs. 18% [177 of 957], P < .001). For genotype 3 patients, the difference in SVR rate was not statistically significant, possibly because of the relatively small sample size for 2003 (51% [78 of 153] vs. 40% [43 of 107], P = .09). In multivariate analysis that included the baseline and treatment initiation variables and was limited to patients starting treatment in 2003, patients receiving PEG-INF 2A remained 41% more likely to achieve an SVR than did patients receiving PEG-INF 2B (odds ratio 1.41, 95% confidence interval [CI] 1.18-1.69, P < .001).

We reestimated the baseline and treatment initiation model for several subsets of the cohort. SVR predictors did not change for the 4,077 patients with race information nor for the 5,241 treatment-naive patients (except prior nonpegylated interferon was obviously no longer relevant). Predictors changed only minimally in analyses limited to the 4,101 patients with a documented follow-up HCV RNA after the end of treatment; recent lifestyle hazards and low-volume facility no longer entered the model. With SVR defined as a negative HCV RNA at least 24 weeks (rather than 12 weeks) after the end of treatment, the overall SVR rate was 24% (1,374 of 5,767), the rate for patients with genotype 1 was 18%, the rate for patients with genotype 2 was 50%, and the rate for patients with genotype 3 was 40%. In multivariate analysis, the significant SVR predictors again did not change. In the baseline and treatment initiation model, the odds ratio of an SVR with PEG-INF 2A treatment compared with treatment with PEG-INF 2B was 1.38 (95% CI 1.19-1.59, P < .001).

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References

Using a large and diverse cohort, we identified several independent predictors of achieving an SVR after PEG-INF/ribavirin treatment. In many previous trials only a few of these factors were identified. This expanded range of predictors may assist clinicians and patients in more accurately assessing the likelihood of an SVR and thus in making more informed treatment decisions.

Our results concur with those of previous trials that identified low-level HCV viremia, absence of cirrhosis, genotype other than genotype 1, and elevated ALT quotient as independent positive predictors of an SVR.4–6, 16 Our results confirm studies that found that African Americans had significantly lower SVR rates than did whites17–19 and that patients who had not responded to a prior course of nonpegylated interferon also had a lower SVR rate.20 Our finding of a decreased SVR rate among patients with diabetes supports recent work indicating that insulin resistance impairs the response to PEG-INF/ribavirin.21

Our results provide new information indicating that the form of PEG-INF may affect the likelihood of an SVR: multivariate analyses adjusted for multiple potential confounding factors showed that patients treated with PEG-INF 2A were nearly 50% more likely to have an SVR. The cause of this apparent difference in efficacy is not clear from the present data and may be multifactorial. PEG-INF 2A and 2B differ in pharmacokinetic properties and HCV RNA clearance kinetics.22–24 The 2 forms also differ in their side-effect profiles.9, 10 Differences in both response while on treatment and side effects may affect continuation of treatment. We did observe a slightly higher discontinuation rate with PEG-INF 2B. When multiple potential confounding variables including treatment duration, which may inadvertently control for some of the mechanism of differential efficacy, were controlled for, patients treated with PEG-INF 2A remained 40% more likely to have an SVR.

The 2 forms also differ in dosing complexity: PEG-INF 2A dosing is fixed, whereas PEG-INF 2B dosing is weight based. The lower response rate for PEG-INF 2B may in part be a result of dosing errors. However, the results of the present and other studies suggest that the acceptable dose range for PEG-INF 2B is much broader than the recommended 1.5 μg/(kg · wk). A dose-finding study for PEG-INF 2B monotherapy showed no relationship between dose and an SVR for a dose above 1.0 μg/(kg · wk),25 and a randomized study found that with ribavirin, PEG-INF 2B at 1.0 μg/(kg · wk) was as effective as 1.5μg/(kg · wk) and better tolerated.26 A flat dose–response curve in this dose range would ameliorate the effect of dosing errors, making them less likely to cause differences in treatment response.

We also identified low baseline cholesterol as an independent negative predictor of an SVR. Low cholesterol may indicate more severe liver disease27, 28 and subsequent reduced treatment response.

Patients treated at low-volume facilities were also less likely to have an SVR in models that controlled for many differences in baseline patient and treatment initiation characteristics. It is possible that other unmeasured characteristics of low- and higher-volume facilities account for this difference. For example, clinician experience or care delivery systems at medium- and high-volume centers might result in an unmeasured aspect of treatment that improves outcomes.

The size of our cohort permitted analysis by genotype, and SVR predictors differed by genotype. Our cohort was 80% genotype 1, and the significant SVR predictors for genotype 1 patients were largely those for the entire cohort. Patients with genotypes 2 or 3 are combined in most studies, but our analyses suggest that genotype 2 patients are more likely to respond to HCV treatment than are genotype 3 patients. In addition, significant SVR predictors differed according to genotypes 2 and 3 and also differed substantially from those for genotype 1. For genotype 2 patients, it may be most important to consider treatment history, BMI, platelet count, and HCV RNA level in making treatment decisions. For genotype 3 patients, none of the patient characteristics considered mattered, but PEG-INF form was important.

Our findings are a reminder that response rates in routine medical practice may be lower than those in clinical trials. Our observed SVR rates were substantially lower than the SVR rates of 42%-52% for genotype 1 and 76%-84% for genotypes 2 and 3 combined that were reported in the 3 largest clinical trials.4–6

The lower response rate resulted in part from including in our population a substantial percentage of patients who would have been excluded from the clinical trials for factors that negatively predict an SVR and a higher percentage of patients with other negative predictors. For example, the 3 pivotal trials required patients to be treatment naive with elevated ALT (ALT quotient > 1) and excluded patients with decompensated liver disease or laboratory contraindications.4–6 Our cohort included patients with previous treatment (12%), normal ALT (32%), decompensated liver disease (4%), and laboratory contraindications (13%). Considering these 4 criteria alone, only 2,640 of our 5,944 patients would have qualified for the clinical trials. Among these 2,640 patients, the SVR rate was 24% for genotype 1, 58% for genotype 2, and 47% for genotype 3, higher than the rates for our entire cohort but lower than those in these clinical trials. These trials also excluded patients with substantial medical or psychiatric conditions (definition of such conditions varied across the 3 trials). Large percentages of our population have diabetes (21%) and a history of depression (49%); many of them might have been excluded from the clinical trials. Our cohort also had a much higher percentage of African Americans (17%) compared with the 3%-5% in the 2 trials that reported data on race,5, 6 a difference that may have contributed to our observed lower SVR rate.

Our lower response rate is also due in part to the much higher treatment discontinuation rates we observed in routine medical practice compared with those found in clinical trials. The 3 trials reported discontinuation rates of 14%-21% for 48-week treatment arms4–6 and 7%-8% for 24-week treatment arms.6 In our cohort, 68% of genotype 1 patients stopped treatment before 48 weeks, and approximately 36% of both genotypes 2 and 3 patients stopped treatment before 24 weeks. The low discontinuation rates in clinical trials are not surprising, as trials generally select extremely motivated participants and have additional personnel to support compliance. In addition, clinical trial investigators and patients generally agree to continue treatment regardless of virologic response while on treatment. In routine medical practice, however, clinicians and patients are not constrained to continue treatment; clinicians may order HCV RNA tests at any time and act on the results. From the available administrative data, we cannot determine the reason for treatment discontinuation. Such a high rate, however, warrants more study of the reasons for treatment discontinuation in our practice and whether similar rates are seen in other practices.

Expansion of the model to include treatment course variables produced little change in the baseline and treatment initiation factors that predicted increased SVR rates. The treatment course variables that entered the model were largely as expected. For example, the chance of an SVR increased with an increase in treatment duration and decreased with a low medication possession ratio for ribavirin. Reductions in the doses of PEG-INF 2A, PEG-INF 2B, or ribavirin during the course of treatment did not enter the expanded model, likely because such dose reductions are typically in response to physiologic effects of the medications, which indicate pharmacologic activity. Receiving epoetin/darbepoetin increased the likelihood of an SVR. This apparent positive effect may also reflect physician response to significant physiologic effects of ribavirin; that is, growth factor use may have been in response to a decrease in hemoglobin. More research is needed to determine the mechanism of the apparent positive effect of epoetin/darbepoetin.

There are several limitations to our findings. This study was based on observational data; patients were not randomized to PEG-INF form or any other aspect of treatment. Although we controlled for many factors that affect the likelihood of an SVR, other unknown and unmeasured factors that correlate with the included variables may account for any observed differences. To determine definitively whether one PEG-INF form in combination with ribavirin is superior for specific genotypes would require a large, randomized head-to-head trial. The ongoing IDEAL trial, which has both PEG-INF 2A and PEG-INF 2B treatment arms, may provide critical information on this issue.29

Patients who previously received nonpegylated interferon and did not respond were less likely to respond to PEG-INF. From the available data, we cannot determine which, if any, patients received prior HCV treatment outside the VA. However, to exclude patients who had started on either nonpegylated interferon or PEG-INF at a non-VA facility and then transferred to VA on HCV treatment, we required patients to have a detectable VA HCV RNA on the closest test prior to their start date for VA PEG-INF. The inclusion of any remaining patients who received nonpegylated interferon outside the VA prior to their VA PEG-INF/ribavirin could contribute to our lower SVR rates.

For several SVR predictors we cannot determine the mechanism or establish causality. For example, differences in viral kinetics, liver iron stores, and immune response have all been proposed as explanations for the lower response rates of African Americans.17 However, we do not have data on any of these factors.

Finally, our cohort consists of United States veterans in VA care, a predominantly male population. Thus, our results may not be generalizable to populations with a larger proportion of women.

In conclusion, with the demonstrated efficacy of PEG-INF/ribavirin against HCV, it is increasingly important to understand the predictors of response to this treatment. Just as SVR rates differ substantially by genotype, so too do the significant SVR predictors. For genotype 1 patients, multiple factors including PEG-INF form apparently affect the SVR rate. Few of these factors affect the rate for genotype 2 patients and even fewer the rate for genotype 3 patients.

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. References