High body mass index an independent risk factor for nonresponse to
antiviral treatment in chronic hepatitis C
Hepatology, September 2003, Volume 38, Number 3
Brian Bressler1, Maha Guindi2, George Tomlinson3, and Jenny Heathcote1
Departments of 1Medicine, 2Pathology, and 3Public Health Sciences, University
of Toronto, Toronto, Ontario, Canada
"...Our data suggest that obesity, only as defined by a BMI greater
than 30 kg/m2 (not weight), is a risk factor for nonresponse to antiviral
therapy independent of genotype and the presence of cirrhosis on pretreatment
liver biopsy in patients with chronic hepatitis C. Obese patients as judged
by their BMI, independent of genotype and cirrhosis, had approximately
an 80% lower chance of a sustained response to therapy compared with normal
or overweight patients. Cirrhosis or HCV genotype 1 were found to be other
independent negative predictors of response to antiviral therapy
"...Previous studies have used weight as a marker to define obesity.
We believe that a more accurate way to separate individuals, according
to their body habitus, is by a measure albeit indirect of total body fat
content. The BMI, which describes relative weight for height, correlates
with total body fat content, whereas weight on its own may not
The aim of this study was to determine if body mass index (BMI) was an
independent predictor of response to antiviral treatment in patients with
chronic hepatitis C.
A retrospective review was performed of all patients at a single center
with chronic hepatitis C treated with antiviral medication from 1989 to
A sustained response was defined as either negative hepatitis C virus
(HCV) RNA by polymerase chain reaction and/or normal alanine aminotransferase
(ALT) level (only in those treated before availability of HCV RNA testing)
6 months following completion of therapy.
All patients were classified into one of 3 groups according to BMI (normal,
<25 kg/m2; overweight, 25-30 kg/m2; obese, >30 kg/m2).
A total of 253 patients were treated with either interferon (IFN) monotherapy
or IFN in combination with ribavirin. Patients were excluded if predetermined
clinical characteristics were unavailable.
Using logistic regression, and after adjusting for the examined variables
(age, sex, history of alcohol consumption >50 g/d, cirrhosis on pretreatment
biopsy, and BMI), likelihood ratio tests showed significant differences
in response to treatment according to BMI group (P = .01), genotype (P
< .01), and cirrhosis (P < .01).
Those with genotypes 2 or 3 had an odds ratio (OR) for success of 11.7
compared with those with genotype 1, cirrhotic patients had an OR of 0.15
compared with noncirrhotic patients, and obese patients had an OR of 0.23
compared with normal and overweight patients.
Hepatic steatosis (fatty liver) was not an independent risk factor for
response to antiviral treatment. In conclusion, obesity, only when defined
as a BMI greater than 30 kg/m2, is an independent (of genotype and cirrhosis)
negative predictor of response to hepatitis C treatment.
The mechanism whereby obesity may affect the antiviral response to treatment
is not completely understood. BMI has been shown to correlate with the
degree of steatosis seen in hepatitis C, and our study confirms this association.
Steatosis leads to an increase in lipid deposits within cells that may
cause a functional disturbance by decreasing the contact area between
the drugs and the hepatocytes containing the virus, thus causing a reduction
in antiviral drug efficacy. Furthermore, the degree of steatosis has been
shown to correlate with the severity of fibrosis. Clouston et al. showed
that this relationship between steatosis and fibrosis in patients with
chronic hepatitis C was significantly associated with steatohepatitis-like
perisinusoidal fibrosis in acinar zone 3.
Our study did not show an association between the presence of greater
than 5% steatosis with sinusoidal fibrosis; however, it did show that
BMI and fibrosis were risk factors for nonresponse but independent of
It has also been shown that HCV genotype 3a is associated with steatosis
independent of body weight. However, individuals infected with genotype
3a, although they may have hepatic steatosis, have an excellent response
to antiviral therapy. In those with genotype 3a, hepatic steatosis disappears
with loss of viremia. The response of patients with genotype 3a to antiviral
treatment indicates that it cannot be hepatic steatosis alone that decreases
the antiviral response. Our study showed that, even though a BMI greater
than 30 kg/m2 predicts the presence of hepatic steatosis, it is only the
BMI that remains an independent risk factor for a poor sustained response
to antiviral treatment. Furthermore, the presence of hepatic steatosis
does not influence a patient's response to antiviral therapy when their
BMI is taken into account.
One study has shown that the pharmacologic effect of IFN on the induction
of 2,5´-oligoadenylatesynthetase, an enzyme induced by IFN, is reduced
in obese patients following an injection of standard IFN. Response to
IFN in obese individuals could cause a reduction in the initial absorption
of drug given subcutaneously because of an increase in subcutaneous fat.
This may be particularly relevant if the injection of IFN is abdominal
because abdominal fat mass measured by waist circumference is associated
with hepatic steatosis, thus compounding nonresponsiveness. Therefore,
there could be multiple mechanisms contributing to the nonresponsiveness
of individuals with large waist circumferences.
The structural property of some pegylated IFNs is such that its size precludes
rapid uptake into the vascular system. Large proteins (>15 kd) injected
subcutaneously are primarily taken up by the lymphatic system. Furthermore,
examining the lymphatic transfer of a series of compounds with increasing
molecular weights has led to the conclusion that there is a direct correlation
between larger-molecular-weight molecules and the greater extent of lymphatic
absorption and reduced volume of distribution. Therefore, standard IFN
is absorbed through blood capillaries and the lymphatic circulation; however,
because of the size of some pegylated IFN, they may be predominately taken
up by the lymphatics. Obese people are known to have poor lymphatic circulation,
and this could potentially lead to lower serum levels of pegylated IFN,
thus diminishing the likelihood of a successful antiviral response.
Our data suggest that BMI but not body weight should be considered when
advising an individual with chronic hepatitis C of the likelihood of a
sustained viral response following a course of antiviral treatment. This
study does not address the issue if weight-based dosing of standard and
pegylated IFN would compensate for the lower response rate in patients
with a BMI greater than 30 kg/m2. Prospective studies are needed to investigate
the response rate in obese patients using higher doses of pegylated IFN.
Furthermore, if the BMI specifically reduces the antiviral response, weight
reduction may enhance antiviral responsiveness in obese patients with
chronic hepatitis C.
Editorial in Hepatology: HCV's resistance to treatment is being nurtured
Arthur J. McCullough, M.D.
Division of Gastroenterology and The Schwartz Center for Metabolism and
Nutrition at MetroHealth Medical Center, Case Western Reserve University,
Bressler et al. are to be commended for their attempts to estimate body
fat mass in 3 different ways: BMI, body surface area, and body weight
by quartiles. However, all 3 of these related measurements are potentially
inaccurate due to fluid retention in cirrhosis even in the absence of
ascites and edema. Therefore, it would have been beneficial if the fluid
status and number of patients with cirrhosis had been reported. Another
estimate of body fat mass, such as anthropometry, would have supplied
additional support for the conclusions based on BMI. In contrast to these
limitations, liver biopsy was used to estimate hepatic fat, a parameter
that is more reliable and whose pathologic interpretation is accurate
and reproducible. However, hepatic fat is not uniformly distributed, and
the possibility of sampling error exists.
The second issue is based on the notion of fat distribution being more
important than the total body fat mass. Visceral fat has been shown to
be a predictor of hepatic steatosis in HCV patients and obesity. Therefore,
waist circumferences (not likely to be available in retrospective study)
would have been of interest to compare with BMI.
Regardless of these considerations, the two important observations in
the current study appear valid. The first is that obesity is an independent
factor that diminishes the eradication rate of IFN-based treatment. This
confirms previous studies that suggested obesity to be an independent
negative predictor of HCV response to treatment. Although the mechanism
by which obesity diminishes the efficacy of antiviral treatment in the
HCV patients remains to be elucidated, there are at least 3 likely possibilities.
(Two were discussed by Bressler). The third possible mechanism is altered
immune function associated with obesity, in part mediated by leptin resistance.
A rigorous Th-1 response stimulated by antiviral treatment improves HCV
clearance with abnormalities in helper T cells and Th1/Th2 ratios being
more common in nonresponders to therapies. Leptin is a pleiotropic molecule
that regulates T-cell response, polarizing Th cells toward a Th1 phenotype.
However, in states of leptin resistance (such as obesity), disordered
T-cell function may prevent viral clearance. It should be emphasized that
leptin resistance is just one aspect in the mosaic of obesity-induced
immunologic abnormalities that may play a role in HCV resistance to treatment.
Although the efficacy of antiviral therapy in chronic hepatitis C has
improved since standard interferon (IFN) monotherapy was introduced, nonresponse
to the current therapies remains common. Several factors have been shown
to influence response; these include viral factors (particularly genotype)
and host factors, which may be genetic (sex, HLA type, cytokine polymorphisms),1-3
and others such as age, presence of cirrhosis, and race.
In patients with chronic hepatitis C, the degree of hepatic steatosis
and fibrosis has been shown by some to correlate with body mass index
(BMI), whereas others report that this is not so. Steatosis may be an
important cofactor in both accelerating fibrosis and increasing liver
necroinflammatory activity in chronic hepatitis C.
Studies show that severe fibrosis or cirrhosis seen on liver biopsy is
associated with a reduced rate of sustained viral response when treatment
with either IFN-2b and ribavirin or pegylated 40-kd IFN-2a monotherapy.
Obesity, a modifiable risk factor, may have a deleterious effect on treatment
response to both pegylated and standard IFN monotherapy. Pharmacodynamic
studies indicate that obesity may be associated with a weaker biologic
response to standard, exogenous IFN-2b. Alternatively, the greater likelihood
of more severe liver disease in obese individuals may be the factor influencing
response to antiviral therapy.
The aim of this study was to identify whether obesity defined according
to BMI is an independent risk factor for impaired response to antiviral
therapy in individuals chronically infected with hepatitis C.
A retrospective review of all patients with chronic hepatitis C at a single
center given antiviral therapy from 1989 to 2000 was conducted.
All patients were classified into one of 3 BMI groupings (normal, <25
kg/m2, overweight, 25-30 kg/m2; obese, >30 kg/m2). Patient weight alone,
categorized in 3 ways, was also used as a variable to determine the use
of this parameter to predict response (bottom one fourth, middle one half,
and top one fourth of weights as 3 categories; weight as a continuous
predictor; and 2 groups with weight <85 kg compared with >85 kg).
Patients were also divided into groups according to their body surface
area. Data on patients were collected for sex, age at initiation of treatment,
previous ethanol consumption (differentiated as >50 g/d or <50 g/d),
hepatitis C virus (HCV) genotype, and evidence of cirrhosis on pretreatment
A single pathologist reviewed biopsy specimens with 2 hematoxylin-eosin-stained
sections for necroinflammatory grading and 2 Masson trichrome-stained
sections for assessment of fibrosis. The histologic variables evaluated
were as follows. (1) Large-droplet steatosis was evaluated by 2 methods.
The first method was a direct estimate of the percentage of hepatocytes
containing large-droplet fat. The second method used the Brunt grading
system for steatosis in which steatosis is graded 0 to 3 based on percentage
of hepatocytes involved (0, none involved; 1, up to 33%; 2, up to 66%;
3, more than 66%).
A total of 253 patients were treated with either IFN monotherapy (either
standard or pegylated) or IFN-2b in combination with ribavirin. Combination
treatment was given to 101 (40%) of the patients. Patients were excluded
if there were missing data in any of the predetermined variables (except
for those treated before availability of genotype testing, who were classified
as genotype unknown); thus, 174 patients fulfilled study criteria. Of
the 174 patients, 48 received combination standard IFN-2b and ribavirin,
96 received standard IFN-2b monotherapy, and 30 patients received pegylated
IFN-2a monotherapy (24 at a weekly dose of 180 µg, one at a weekly
dose of 135 µg, and 5 at a weekly dose of 90 µg). Sustained
response was achieved by 73 patients (42%); 43 patient responses were
confirmed sustained virologic responses. Thirty percent of the 174 patients
were female, the mean age of the patients was 44.9 ± 8.6 years,
93 of 174 (53.4%) were genotype 1 (in those in whom genotype was tested),
25 (14.4%) had a history of ethanol consumption greater than 50 g/d, and
42 (24%) had cirrhosis on pretreatment biopsy. Seventy patients (40.2%)
had a normal BMI (<25 kg/m2), 69 patients (39.7%) were overweight (25-30
kg/m2), and 35 patients (20.1%) were obese (>30 kg/m2). Mean weight
of patients was 78 kg.
In a logistic regression model that included sex, ethanol consumption,
viral genotype, cirrhosis on baseline liver biopsy, patient weight according
to the 3 categories mentioned, and BMI group, likelihood ratio tests showed
significant differences in response to treatment according to BMI group
(P = .01), genotype (P < .01), and cirrhosis (P < .01). In particular,
those with genotypes 2 or 3 had an odds ratio (OR) for a sustained response
of 11.6 compared with genotype 1, patients with cirrhosis had an OR of
0.15 compared with those without cirrhosis, and obese patients (BMI >30
kg/m2) had an OR of 0.24 compared with normal and overweight patients.
The response to treatment was not predicted by a patient being overweight
(BMI, 25-30 kg/m2) (OR, 1.7; 95% CI, 0.68-4.2). The probability of a sustained
response based on viral genotype, BMI, and liver histology has been calculated.
Those individuals who were obese, cirrhotic, and infected with genotype
1 had a predicted probability of 0.013 (95% CI, 0.0023-0.069) for a sustained
response to therapy. Weight on its own rather than calculated BMI as an
indicator of obesity did not predict response. There was no pattern of
association between treatment response and weight alone evaluated by 3
methods (bottom one fourth, middle one half, and top one fourth of weights
as 3 categories; weight as a continuous predictor; and 2 groups with weight
<85 kg compared with >85 kg). When body surface area was used instead
of BMI in the logistic regression model, increasing body surface area
shows a trend to lower success with antiviral treatment (P = .08). However,
when body surface area is added to the logistic regression model, which
included BMI, body surface area is not a significant factor. On the other
hand, with both BMI and body surface area in the same model, BMI remains
a significant factor (P = .04).
Evaluation of cohort with liver histology available for reevaluation
Of the 174 patients who were included in our study, 136 had liver biopsy
specimens for reevaluation. From this subgroup of patients, 29 had a BMI
greater than 30 kg/m2 and 107 patients had a BMI less than or equal to
30 kg/m2. Of the 29 obese patients, 27 (93%) had a steatosis grade of
1 or more; of the 107 patients who were either overweight or had a BMI
less than 25 kg/m2, 67 (63%) had a steatosis grade of 1 or more. The grade
of steatosis was statistically greater in the obese group compared with
the overweight and normal patients (P < .005). Of the 29 obese patients
with BMI of >30 kg/m2, 3 (10%) had steatosis grade 2. The presence
of greater than 5% large-droplet steatosis was not significantly associated
with sinusoidal fibrosis (P = .16). In a logistic regression model that
included those variables that were independent predictors of a poor sustained
response to therapy in our original cohort (viral genotype, cirrhosis
on baseline liver biopsy, and a BMI >30 kg/m2) and steatosis grade
(Brunt grading system), likelihood ratio tests showed a BMI greater than
30 kg/m2 to be a significant predictor of response to treatment (P = .05).
Steatosis grade of 1 or more was not a statistically significant predictor
of response in this model.