Patients with hepatitis C have two concerns. The first concern relates to how likely it is that he or she will develop end-stage liver disease, namely either symptoms of cirrhosis that will incapacitate them, lead to the need for a liver transplantation, or even death, or primary liver cancer (hepatocellular carcinoma). The second concern is what can be done to prevent, or at least reduce the likelihood of, the development of those conditions. The first relates to the natural history of the process and the second relates to the benefit of treatment.
Before discussing these issues, let us consider the difference between signs and symptoms. Signs are typically abnormal blood tests; these are the tests that doctors order to monitor patients. Some signs measure the degree of inflammation (such as enzymes called aminotransferases, known by patients as “ALT” or “AST”). While the ALT or AST test is often called a “liver function test”, it really does not measure any function that the liver performs, but simply assesses the presence of liver cell damage. It is more appropriately denoted as a “liver enzyme test”.
The other test (sign) that we need to consider is the one that measures the number of hepatitis C particles that are present in the blood (known as “viral load” or “viral titer”). Traditionally, the success of treatment has been measured by the absence of detectable viral particles in the blood at least six months after the therapy has been stopped. This prolonged clearance of virus from the blood stream is known as a “sustained viral response” or SVR.
Symptoms are the things that affect the quality and quantity of life; it is the development of symptoms that concerns patients. Symptoms of liver failure include internal bleeding from dilated veins (called varices), excessive fluid accumulation in the legs (called edema) and abdomen (called ascites), and slowing down of thinking with confusion and even coma (called hepatic encephalopathy). Another symptomatic complication of cirrhosis is the development of hepatocellular carcinoma.
The point is that if only the enzyme tests remain abnormal, and symptoms of liver failure never occur, there is no clinical “dis-ease”. Treatment is irrelevant because it is impossible to make an asymptomatic patient feel better and, at the time of treatment, patients do not have symptoms of end-stage liver disease. However, as alluded to above, physicians use signs to assess success or failure of treatment. When signs are used to measure success, we call them “surrogate” outcomes. Physicians are operating under the assumption that, if the treatment makes the surrogate outcomes better, the patient will not get symptoms of liver failure in the future. We will return to the use of surrogate outcomes shortly.
The term “natural history” refers to what the course of a disease would be if no intervention occurred. While a number of studies have assessed the natural history of hepatitis C infections, most of them suffer from a methodologic defect known as selection bias. For example, suppose you work at a liver transplant center and only see patients who are referred there to have such a procedure. If I were to ask you what the natural history of hepatitis C is in your experience, you would say that all of the patients that you see have end-stage disease. On the other hand, if you work at a blood donor facility and I asked you the same question, you would say that all of the hepatitis C positive people that you see are quite healthy (since, if they were not, they would not be donating blood). These examples reflect the different populations who are seen in special centers, but neither is representative of all people infected with hepatitis C.
Most natural history studies employed data collected in centers that treat liver disease, where sicker hepatitis C patients tend to congregate. While data from such studies have suggested that it takes, on average, about 20 years to develop cirrhosis and 30 to develop hepatocellular carcinoma in patients who are infected with hepatitis C (1, 2), that observation is confined to the patients referred to that center who developed cirrhosis or cancer. It is incorrect to assume that all patients who are infected with hepatitis C progress to these end points in two or three decades. In order to calculate the probability that end-stage liver disease will occur, one needs to know the denominator, namely the entire number of people who got infected, in addition to the numerator, namely how many of those people developed end-stage liver disease.
Several research centers actually identified cohorts of people (a cohort is the denominator) when they first got infected and have followed those people for decades. Most of these studies included people who were transfused in the 1970's and, within months afterward, were found to have developed post-transfusion hepatitis C. Decades later, no more than 10% (and usually far fewer) of these infected cohorts had end-stage liver disease (3-10).
Population statistics also attest to the fact that the vast majority of infected patients will never get into trouble. It has been estimated that there are 3-4 million people in the United States who are infected with hepatitis C. The number of patients dying (or requiring liver transplants) each year as a result of hepatitis C infection is no more than 10-20,000. If you had 3 million socks in a drawer and took 20,000 out each year, it would take you 150 years to empty the drawer. (If you had 4 million socks and took out 10,000 each year, it would take 400 years to empty that drawer.) Since it is only a matter of a few years between the time that patients develop symptoms of end-stage liver disease and die of, or are transplanted for, that condition, and since the vast majority of patients become infected after age 20, most patients infected with hepatitis C will have to die of something else before their livers fail.
This is not to say that hepatitis C never progresses to symptomatic end-stage disease or hepatocellular carcinoma. It is to say that, going in, a particular patient’s likelihood of having that happen is no more than 10% or so. The reason that hepatitis C is the leading cause for liver transplantation is not that progression occurs frequently, but rather that even a small fraction of a large number is a relatively large number.
It should also be appreciated that hepatitis C infection is not analogous to infection with the human immunodeficiency virus (HIV). Patients infected with HIV, if left untreated, will almost universally die of some HIV-related disease.
Even if the probability of developing end-stage liver disease is low, it is not zero. Thus, treatment might still be a consideration if it were to be demonstrated that the probability of developing end-stage disease of hepatocellular carcinoma was made lower as a result.
The best way to prove that a treatment is effective at preventing morbidity or mortality is to compare individuals who were treated with those who were not (the control group). Ideally, the two patient groups should be identical in all other respects except the receipt or non-receipt of the treatment.
To make this point with regard to the treatment of hepatitis C, there have been some studies published in which investigators at liver centers compared the patients who had been treated to those who had not and have claimed that the treated group had better outcomes. However, when the individual patients were assessed, it turned out that the patients who did not receive treatment had a reason for not receiving the treatment, usually that they were too sick to be treated. This difference becomes what is called a “confounding factor” and makes it impossible for us to know if it was the treatment or the underlying worse disease that was responsible for the different outcomes.
The scientific way to get around confounding factors is to assign patients to one or the other group randomly; such a study is known as a randomized clinical trial. Unfortunately, since it takes decades for patients to develop the symptoms of end-stage disease or cancer, it has been believed to be impractical to undertake such trials in general. Instead, researchers have relied on the surrogate outcomes. Initially the enzyme tests were largely used, but more recently, perhaps mimicking how HIV infections are treated, the endpoint of the randomized trials has been the SVR.
Originally, hepatitis C was treated just with interferon (IFN) injections, usually three times per week. The SVR rates were generally low, somewhere around 15-20% (compared to 0-1% in the untreated control groups). Subsequently, other drugs have been employed; the second was ribavirin (which was used in combination with IFN), followed by pegylated IFN (Peg-IFN), and most recently by the protease inhibitors telaprevir and boceprevir. Combination treatment with some or all of these medications is now able to achieve SVR rates in excess of 60%.
These medications have all been approved by the Food and Drug Administration because of the effect that they had on the surrogate outcomes. However, in order to be sure that a surrogate outcome is an appropriate one on which to base research trials and clinical decisions, that surrogate outcome has to be validated. In other words, it has to be shown that the surrogate and clinical outcomes change similarly when treatment is provided. Because of the problem with confounding factors, such a demonstration must occur in randomized trials.
The most-cited reason for accepting SVRs as surrogate outcomes is the fact that, if one looks at patients who obtain them, they have better longer-term courses than do patients who fail to achieve them. Of course, the problem with such an argument is that the patients who did not achieve them also were treated.
There are reasons to be concerned that SVRs are not an adequate surrogate outcome. First of all, a small minority of patients have been reported to develop end-stage liver disease or hepatocellular carcinoma in spite of having an SVR. Perhaps more importantly, patient factors that are associated with SVRs include absence of cirrhosis or severe fibrosis on the biopsy, female gender, normal body weight, and a relatively recent history of HCV infection (11, 12). These factors are characteristics of individuals who were less likely to get into trouble anyway. If treatment responders are derived from the group destined never to develop end-stage liver disease, than one would expect them to have a better long term course than the nonresponders, even if no treatment had been provided.
It is also incorrect to use the HIV model as justification for the use of SVRs as a surrogate outcome. The titer of HIV does predict the clinical course in HIV-infected individuals, and viral clearance has been validated as a surrogate outcome. It could be postulated that, if the serum is the means whereby the HIV reaches its target cell, the lymphocyte, the serum level is an important factor in the progression of AIDS. However, we have known for decades that the titer of hepatitis C virus is not predictive of the clinical course. This may be because the titer is not important with regard to the virus infecting other liver cells; such infections do not need the bloodstream to act as a carrier, since the virus can simply infect neighboring liver cells directly. Unlike the HIV titer, no effort has been made to validate the SVR.
What would be needed are long-term randomized trials assessing the ability of anti-viral therapy to prevent the development of end-stage liver disease in patients with hepatitis C. Such studies would include untreated control groups and would be conducted long enough to see clinical events in addition to the surrogate outcomes. For the reasons noted, such trials have not been conducted in the average risk patient. However, since patients who already have severe fibrosis are more likely to develop clinical events in the nearer future, such patients have been included in five small treatment trials (13-17). All of these trials randomized patients into one of two arms, either treatment with standard IFN alone or no treatment. For the most part, no effect on the clinical outcomes was observed. In fact, the single trial that described a benefit (14) was only described in a preliminary form as an abstract for a meeting in 1999; for reasons that are not clear, this trial was never published as a full paper in a peer-reviewed medical journal in the subsequent 13 years.
More recently, another three trials that randomized patients with severe fibrosis (usually cirrhosis) into a treatment arm (Peg-IFN) or no treatment have become available. Most of these patients had failed a prior course of combination therapy and were given the Peg-IFN for a protracted period of time. The results of these trials have been described in a systematic review that my colleagues and I recently published in the Cochrane Library (18) and the findings are quite concerning.
When the three trials were put together using a widely-accepted statistical technique known as meta-analysis, the treatment did not provide any favorable clinically relevant benefits. In fact, when the two trials that were believed to be based on the best scientific methodology were considered, the recipients of the Peg-IFN had an increased mortality compared to the controls. While this could simply be a statistical chance event, it is the only information that we currently have and, at the very least, future patients who are going to be advised to have treatment with interferon-based combination therapy need to know about this observation as part of the informed consent process.
Peg-IFN also caused a number of adverse events, including infections, flu-like symptoms, fatigue, muscle aches, fever, rash, and lowering of white blood cell and platelet counts.
An even more troubling observation of this review was that, in spite of the patients not doing better with regard to clinical outcomes (and even doing worse with regard to side effects and, perhaps, mortality), the surrogate outcomes (SVR in particular, but also markers of inflammation such as the ALT) did significantly improve. In other words, the surrogate markers failed to validate. This means that, going forward, we have no valid outcome to use to measure success of treatment. While these surrogate outcomes may work in other scenarios (i.e., other than a patient with severe fibrosis who is being retreated with Peg-IFN alone), we require proof of that speculation; the surrogate will have to be shown to be valid, which will require a randomized trial that does assess clinical and surrogate outcomes.
These antiviral drugs are not free of medical or financial costs. IFN has to be injected. It causes flu-like symptoms (usually mild) in almost everyone who takes it. At least 5-10% of treated patients develop such significant problems (e.g., a more severe fatigue or bone marrow suppression) that they have to discontinue the treatment. Clinically important and irreversible thyroid problems have been induced. Some patients can have a transient increase in their markers of inflammation (e.g., ALT); a few have had severe, and even fulminant, hepatitis. IFN may accentuate underlying depression or anxiety. There have been fatalities reported, from exacerbations of hepatitis or other immune-mediated diseases (since interferon stimulates the immune system) and from suicides.
Ribavirin causes a hemolytic anemia in almost all recipients, presenting a potential problem particularly in patients with arterial disease (and compromised blood flow to various tissues). There is also concern about its potential teratogenicity. The protease inhibitors also have significant side effects, including anemia and skin rash.
A one-year course of combination Peg-IFN/ribavirin treatment will cost more than $20,000. An estimate in the Veterans Affairs health care system was that, if all veterans were to be screened in order to identify the carriers, and if 20% of the hepatitis C carriers were eligible for treatment, the cost would be in excess of $600,000,000 (19).
Given the natural history of chronic hepatitis C, as well as what we know therapy accomplishes, it is very difficult to justify a policy for routinely treating such patients to prevent decompensated liver disease. The surrogate outcomes were not valid in the one occasion when validation information was available. The treatment has not been proven to be efficacious with regard to preventing clinically important disease, it is expensive, and it causes substantial morbidity (including death). It is an inappropriate clinical decision to prescribe a toxic therapy (especially an expensive one) that has never been shown to provide clinical benefit in properly-done randomized trials.