… In Non-symptomatic, Non-Bulky, Previously Untreated Follicular Lymphoma … Commenting on A Preliminary Analysis *
Greetings,
The recent ASH conference led to many press releases with titles like “Rituxan delays chemo for slow-growing lymphoma” and “Roche’s MabThera Delays Need for Additional Lymphoma Treatment, Study Says. It’s our impression that the importance of the study has been vastly overblown in the flurry of media hype surrounding the conference.
Within we discuss the limitations of Progression Free Suvival and Time to New Treatment as ways to estimate clinical benefit in this study, and explain why longer follow up is needed to tell if either approach provides meaningful clinical benefit.
TIME TO NEW TREATMENT AS PRIMARY ENDPOINT?
The abstract published at ASH (see link below) shows that Rituxan, when administered as first therapy – in this low risk and previously untreated population — delays the initiation of new treatment.
The authors propose that “Time to New Treatment” is a way to compare Rituxan to Observation in previously untreated follicular lymphoma
… the use of the word “NEW” being a creative way to compare very different events. That is, “New treatment” means: Time to FIRST Treatment for the observation group, versus Time to SECOND Treatmentin each of the Rituxan-treated arms of the study, because, stating the obvious, observation is not a treatment.
The abstract doesn’t provide information on what clinical changes triggered the decisions to initiate the NEW therapy in the respective groups, nor does it explain why chemotherapy is the overwhelming new therapy choice, even in the observation group.
… The selection of new therapy (chemo – 84%) suggests an expectation among these investigators that this use of Rituxan is appropriate only in patients with low levels of disease – judged to be not in need of treatment. Is there clinical trial data that validates this expectation?
“The primary endpoints were a) time to initiation of new therapy (chemotherapy or radiotherapy) and b) effect on quality of life.”
“New treatment was chemotherapy in 78 (84%),
radiotherapy in 10 (11%),
rituximab monotherapy in 2 (2%),
surgery in 1(1%),
currently not known in 2 (2%).”
Be aware that when to treat can be a very subjective decision for the indolent lymphomas, for which there is a wide range of indications for treatment, including “patient preference.” ( See NCCN guidelines http://lymphomation.org/prognosis.htm#NCCN )
… That is, even for the same patient, the decision to treat can made long before, or a long after, treatment is indicated by NCCN or GELF criteria — a decision that is dependent on patient reporting of symptoms, as well as physician and patient preferences.
That Time to Next /New Treatment is a fuzzy endpoint is a familar concept to patients with indolent lymphoma who have consulted multiple oncologists about when to treat.
Since the investigators were aware which patients were receiving therapy or observation (it was an open label design), these were not blinded decisions – and therefore could be influenced (even if unconsciously) by investigator bias – and perhaps the patients discomfort with observation, which is often described by them as “watch and worry.”
So let’s take a look at some of the data pulled from the abstract, with a focus on the observation (“watch and wait”) group:
The estimated median time to initiation of new therapy in the observation arm (arm A) was 33 months.
So the participants randomized to observation received what might be called a benefit of 3 years of therapy without treatment risks – again, assuming Time to NEW Treatment is a reliable measure of clinical benefit.
Arm A – observation:
Spontaneous remission was seen in 3%
Partial Response =6%
NC (no change) =74%
Progressive disease PD=17% *
* An obvious follow-up question on the incidence of PD being: What percent of the participants in the observation group were treated despite having had no progressive disease (PD), and how did that compare with arms B and C?
… Progressive Disease (PD) seems the least controversial trigger leading to treatment (compared to patient-reported symptoms), assuming its validated by imaging or some other objective measure.
… Since 17% in the observation arm of the study had PD it would be informative to know what were the other clinical events that led to treatment in the respective arms of the study - noting that the graphic shows more than 50% needing treatment at 48 months in the observation arm, and that 74% in the observation arm were reported to experence No Change (NC) in the status of the lymphoma.
Versus the Rituxan treatment groups:
Arm B: CR+CRu=45%, PR=33%, NC=19%, PD=3%.
Arm C (with maintenance): CR+CRu=49%, PR=36%, NC=11%, PD=3%.
EFFECT ON QUALTITY OF LIFE (QOL)?
There is no compelling information provided on this measure, despite it being listed as a primary endpoint in this study. QOL is very difficult to measure and compare objectively. However, adverse drug events can be measured, and, obviously, favor those who were not treated. While drug-related events might be offset by decreased symptoms in the Rituan-treated groups, such QOL measures were not provided:
Drug-related Serious Adverse Events (SAE):
(SAE were considered possibly, probably or definitely related to the study drug)
Observation arm = 0/186
Rituxan arm =4/84,
Rituxan maintenance arm = 10/192).
5 allergic reactions (two grade 3), 6 infections, 3 episodes of grade 4 neutropenia.
SURVIVAL DIFFERENCES?
Survival is the most reliable endpoint for determining if either approach provides clinical benefit. The other measures (such as PFS) are called surrogates which depending on the study design, the population, and the magnitude of the difference, may reasonably predict a surival benefit … or not.
… Here we see no difference in mortality comparing the three treatment arms at this time. That the sponsor has chosen Time to Next Treatment as the primary endpoint doesn’t mean that it it’s a good choice of endpoints to predicts a survival benefit. That Survival was not listed as a primary endpoint raises the concern that follow-up of the participants might not be extended to compare survival outcomes in these patients.
Survival, all groups: “98% of patients remain alive and there was no different in overall survival between the 3 arms (p value >0.5).”
PROGRESSION FREE SURVIVAL?
Progression Free Survival (PFS) is a combination endpoint measuring both mortality AND progression events and counting each as one event.
… To a lay audience PFS probably suggests a survival benefit, but in this case it has nothing to do with survival because there was no difference in survival between the observation and treatment groups in this study.
Again: “98% of patients remain alive and there was no different in overall survival between the 3 arms (p value >0.5).”
Here we propose – to avoid media intrepretation errors - that PFS not be used in such reporting when there is no difference in the Survival part of this composite endpoint.
When reviewing commentary on this study, be aware that the sponsor of the drug will be inclined to present the results in ways that favor more usage of their product, that being the charter of any for-profit organization … thus, it may provide the media with selective comments that support that goal and will tend to understate or fail to adequately discuss the limitations of the study design.
My hope is that the investigators will continue to monitor the participants of this study and report on the survival data and that patients and treating physicians will recognize the limitations of the information provided at this time, and the need for independent discussion of what this data does and doesn’t mean.
Regarding the question whether these results should change clinical practice, we might restate the question this way: Should John and Mary – each recently diagnosed with follicular lymphoma and without symptoms, or bulky disease, or evidence of progression –initiate Rituxan therapy without making use of these clinical indicators?
… The data shows that there are some potential benefits (assuming the benefit can be reliably measured by Time to New Treatment), but also that there was a increased incidence of adverse drug evens (AES) and no survival advantage at this time. Not directly addressed in the abstract is the possibility of poorer responses to subsequent therapy (which is likely to include Rituxan) in this unique population.
So, and as already noted by some experts, only a survival benefit would be persuasive and ought to be practice changing — noting for the benefit of patients that the survival endpoint is persuasive because it accounts for all factors, known and unknown. For mortality events, unlike Time to New Treatment, the meaning would be clear.
Critiques aside, this is a study design that could actually help to answer the question – with longer follow up because patient selection (to observation or Rituxan monotherapy) was randomized; however, the study is unable to answer the key question because of the reliance on questionable surrogate endpoints that do not reliably predict a survival benefit.
Karl Schwartz
President, Patients Against Lymphoma
lymphomation.org
REFERENCE:
* ASH abstract: http://ash.confex.com/ash/2010/webprogram/Paper27692.html