Benefit of brentuximab over bleomycin in first-line treatment of advanced-stage Hodgkin lymphoma has not been proven

TO THE EDITOR:

A randomized controlled trial by Connors et al¹  that was recently published in the New England Journal of Medicine aimed to determine whether brentuximab in combination with doxorubicin, vinblastine, and dacarbazine (A+AVD, n = 664) is superior to standard doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD, n = 670) in treating first-line stage III or IV Hodgkin lymphoma patients. The primary outcome measure was “modified” progression-free survival (PFS), which included progression of disease, death, and second-line therapy application because of a positive end-of-treatment ¹⁸F-fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET) scan as events. The “modified” PFS was determined both by the investigator and by an independent review committee. According to the independent review committee, the 2-year “modified” PFS was 82.1% for A+AVD and 77.2% for ABVD with a risk difference of 4.9%. According to the investigator, the 2-year “modified” PFS was 81.0% for A+AVD and 74.4% for ABVD, with a risk difference of 6.6%. There was no significant difference in 2-year overall survival rate (96.6% for A+AVD vs 94.2% for ABVD, P = .20). Connors et al¹  concluded A+AVD to have superior efficacy over ABVD in treating first-line advanced stage Hodgkin lymphoma patients.

However, we disagree with their conclusion. First, we want to underline that the study by Connors et al¹  was funded and partly performed by the drug industry (Millennium Pharmaceuticals and Seattle Genetics), who had a role in study design, data verification, statistical analysis, and data interpretation. The article was partly written by a medical writer sponsored by the drug industry. These facts emphasize that extra effort should be taken to generate a clear, nonbiased methodology in study design, in order to eliminate biased outcomes due to potential conflicts of interest, particularly considering the very high costs associated with the application of brentuximab.

The validity of the study by Connors et al,¹  however, is heavily biased due to the lack of criteria to define a valid and reproducible outcome measure. Connors et al¹  incorrectly claim that it is well accepted to initiate second-line chemotherapy or radiation therapy on the basis of end-of-treatment FDG-PET results by citing 3 studies to support their statement.^2-4  The study by Barnes et al, however, reported a 4-year PFS of 54% for patients with positive FDG-PET results after treatment. This means that more than half of FDG-PET results after treatment were false positive.²  The study by Engert et al³  included 182 patients with FDG-avid lesions after 6 to 8 cycles of BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone). These patients with “BEACOPP-resistant disease” according to end-of-treatment FDG-PET had an excellent 5-year PFS of 86.2% after the application of radiation therapy only, which strongly suggest false-positive results to commonly occur.³  Finally, the study by Spaepen et al,⁴  which was performed in 2001, included only 5 patients with residual FDG-avid lesions after treatment, which predicted a relapse in all cases.

In contradiction to the claim by Connors et al,¹  current guidelines recommend biopsy of residual FDG-avid lesions after treatment before initiation of second-line therapies.⁵  This statement is supported by recent studies showing high false-positive rates of residual FDG-avid lesions after treatment of Hodgkin lymphoma,^6-8  with false-positive rates up to 66%, with the majority of cases of therapy-induced inflammation being responsible for the residual FDG avidity. It is important to realize that Hodgkin lymphoma comprises only 0.1% to 1% of malignant Reed-Sternberg cells.⁹  As a result, virtually all FDG avidity is caused by surrounding inflammatory cells, even before therapy has been initiated. Because FDG-PET cannot always distinguish between treatment-induced inflammation and residual lymphoma, FDG-PET results after treatment do not always adequately reflect the disease status and cannot be used as a reliable outcome measure or justify second-line therapy initiation.

Another important issue to consider is that, when FDG-PET results after treatment are used determine disease status without histological verification, these results are not reproducible. This issue is reflected by the results of Connors et al¹  (their Figure 1) themselves. Note that their article reported differences between “modified” PFS analyses of the independent review committee and the investigator, despite that fact that the analyses were performed in the same group of Hodgkin lymphoma patients, which emphasizes that expert groups do frequently disagree in determining which lymphoma patient has been cured and which patient has not. The independent review committee reported 117 events in patients treated with A+AVD (90 with disease progression, 18 deaths, and 9 with subsequent treatment after a positive FDG-PET result), whereas the investigator reported 123 events in the same patient group (73 with disease progression, 15 deaths, 35 with subsequent treatment after a positive FDG-PET result). The reason the independent review committee found 3 more deaths than the investigator, which is a quite remarkable finding, remains unknown. Note that the number of deaths reported in the abstract and main text is again different (28 for A+AVD vs 39 for ABVD) than those reported in Connors et al’s Figure 1. The independent review committee reported 146 events in the ABVD group (102 with disease progression, 22 deaths, and 22 with subsequent treatment after a positive FDG-PET result), whereas the investigator reported 164 events in this group (103 with disease progression, 22 deaths, 39 with subsequent treatment after a positive FDG-PET result). The difference between the independent review committee and the investigator in defining an event is reflected in the difference in 2-year PFSs, which were 82.1% and 81.0% (difference of 1.1%) for A+AVD and 77.2% and 74.4% (difference of 2.8%) for ABVD. Note that these differences are actually large considering the fact that the reported differences in PFS between A+AVD- and ABVD-treated patients were 4.9% and 6.6% according to the independent review committee and the investigator, respectively.

In conclusion, the gain in “modified” PFS of A+AVD over standard ABVD may have been highly influenced by the incorrect methodology in the study by Connors et al,¹  which used an outcome measure that was not reproducible and included FDG-PET results, which frequently generate false-positive results. It is of utmost importance that this information is clear to the clinical audience, before a well-balanced judgment can be made on the benefit of A+AVD, where aspects such as PFS, side effects, and costs should be taken into account. Furthermore, rectification of the unjustified claim by Connors et al (who promote to initiate second-line therapies on the basis of a positive FDG-PET result after treatment without histological verification) is crucial to prevent large-scale initiation of highly toxic second-line therapies in patients who are actually cured.