Idenix Pharmaceuticals (IDIX)

[dewophile]

entecavir vs telbivudine

precious - thanks for posting the article. We've already discussed elsewhere the pitfalls of comparing
2 drugs without a direct head to head, but the biases and inaccuracies in this analysis deserve some specific mention and context:

1. first, look at who did the analysis in this article:

"Analysing clinical issues for drugs for hepatitis B
is Jules Dienstag, M.D.,"

I noticed that he works at Mass General with Jules Levin, who was intimately involved in the entecavir trials and presented the 2 and 3 year data at AASLD. coincidence? perhaps..

2. at 1 year he admits entecavir and telbivudine performed similarly. He also claims that although a direct head-to-head was not preformed, the response in the lamivudine control and baseline characteristics point to similar populations. This is flawed as there are many other variables, including viral genotype, that were unaccounted for. Furthermore 1-year lamivudine data is not all that telling as a common control, as efficacy tends to drop off more rapidly in later timepoints (when resistance emerges)..so durability of response to lamivudine would be a better measure, and 2 year lamivudine results were not the same between the two groups as I'll pont out below

3. I won't go into the resistance argument - DD did that nicely in an earlier post. At the end of the
day the main goal of treatment is long-term suppression of HBV DNA, which is directly correlated
to liver inflammation and subsequent disease, and resistance is really only a surrogate marker of risk
of failing to respond to a therapy over time, so lets go to the 2 year data

4. The main argument this author uses to describe how telbivudine doesn't quite measure up to entecavir is the degree of HBV DNA suppression at 2 years in e-Ag positive patients:

"for example, for HBeAg-positive patients, at 2 years, HBV DNA was undetectable in 87% of an entecavir group compared with 54% of a telbivudine group"

Actually it was 57% for telbivudine on a strict ITT basis (http://www.natap.org/2006/AASLD/AASLD_18.htm), and 80% (NOT 87%) cumulative undetectable rates for entecavir in the (ETV-22) study, and mind you this was
NOT on an ITT basis. The 87% figure comes from the 3 year rollover extension trial, that only included patients who were eligible for a 3rd year of entecavir therapy based on virologic and serologic status - which amounted to 119 patients of the original 354 in the ETV-22 study..so among this smaller subgroup rolled over to the ETV-901 3 year analysis, when you
look back 85% happened to be HBV DNA negative at 2 years (http://www.natap.org/2006/AASLD/AASLD_19.htm)...again, this is not much different than the 80% cumultive 2-year undetectable rate, but I point it out only to show what is a glaring inaccuracy, either due to bias or simple carelessness on the part of this reviewer

so how to reconcile the 80% vs 57% 2 year results among hbeAg+ patients. Well firstly, there is a reason why JP sommadossi stresses the ITT issue in the GLOBE analysis (and lack thereof in other HBV studies)..in the ETV-22 study entecavir was discontinued at 1 year in patients who had a "response", defined as HBV DNA <0.7 MEqmL by bDNA & HBeAg loss..it is not clear to me how these 74 patients (out of 354) were included in the analysis..nor is it clear how the dropouts were handled. If the 74 were counted as negatives based on 1-year results - which is intimated here as the authors describe "cumulative" HBV DNA-negativity at 2 years - then the data is likely skewed upward as a proportion of these negatives would likely drift above the undetectable range at year 2. In fact, there is a table in the 022 study that describes a 67% HBV undetectable rate at 96 weeks, which I think includes 2-year results among all patients, including those who stopped therapy at 1 year - this seems to me the closest measure of a true ITT than the 80% “cumulative” rate of patients who ever achieved HBV undetectability described in the text (http://www.natap.org/2005/AASLD/aasld_37.htm).

Secondly, there is a signal here indicating the populations were NOT equal. 2-year data in e antigen
negative patients shows very different response to lamivudine controls - in the GLOBE trial only 57% were undetectable versus 77% in the entecavir 027 trial. So were the 2 populations truly equal? why did the lamivudine controls in the entecavir 027 trial, which were recruited at the same sites as the 022 trial, fare better than those in the GLOBE trial at 2 years among the e Ag negatives? were HBV genotypes distributed differently with a greater proportion of more virulent and treatment-refractory patients in the telbivudine trial - a variable that was NOT controlled for in the trials?
now among e Ag + patients, lamivudine controls did perform identically in the GLOBE and entecavir trials (39% undetectable in both trials)..so why the difference in lamivudine responses between the e+ and e- patients? both groups were recruited from same centers - so variables such as genotype, which vary based largely on geography, should be similar? the answer is I have no idea..it could be that a response to lamivudine at 2 years in this subset is simply not adequate to give a signal as to the virulence of the HBV strains? perhaps the difference in statistical analysis between the trials could obfuscate the data? or maybe the e Ag + patients were very similar between the 2 trials somehow, but not the eAg- ..who knows, but my point is we really don't have the answers and to me this again highlights the pitfalls of comparing 2 drugs from separate trials in separate populations using different statistical methodology

5. as long as we're talking about durability of response, for the HBeAg negative population at 2 years the data was as follows:

entecavir 94% HBV undetectable...lamivudine 77%
telbivudine 82% undetectable......lamivudine 57%

the incremental improvement over the lamivudine control is actually higher in the telbivudine trial..i point this out not because I feel this is a valid comparison either, but I just want to illustrate that manipulating the data bewteen 2 seperate trials can cut both ways if we're going to do some statistical gymnastics here with the data

6. Lastly, as long as the 2 year data specifically in the eAG+ population is what is being used as an
argument in favor of entecavir, how about we look at another important endpoint in this population at 2 years - e Ag seroconversion:

HBeAg seroconversion at 2 years in the GLOBE trial was 30% for telbivudine and 25% for lamivudine. in the 022 study it was 31% cumulative rate for entecavir and 26% cumulative for lamivudine

what about LFT normalization?:
telbivudine 78%, entecavir 79%

so is there a clinically meaningful difference between the 2 agents at 2 years in eAG+ patients?


In conclusion, we really don't know how these 2 agents compare head to head until data from a head to head emerges. viral kinetics from both drugs is reasonably similar at most timepoints, and there is some signal to indicate the populations may be dissimilar based on 2-year response rates to lamivudine among e-Ag neagive patients. In addition, the differences in the statistical analysis in the trials further clouds any comparison. what we do know, and the only thing we really do know, is that both agents perform well, both have superior efficacy to lamivudine, and both are more potent than adefovir as monotherapies...that's about all we know for sure, and anyone trying to argue otherwise is speculating, conjecturing, reading into and twisting the data, and is basically violating an important tenet of proper scientific method and statistical analysis