Most guidelines for Barrett’s esophagus (BE) recommend radiofrequency ablation (RFA)
as first-choice option for dysplastic BE, or after endoscopic resection of early cancer.
Numerous high-quality (randomized) studies have demonstrated that RFA for these indications
is efficient and safe [1]
[2]. In recent years, an increasing number of studies appeared, evaluating use of new
ablation techniques such as cryospray ablation, cryoballoon ablation (CBA) and (hybrid)
argon plasma coagulation (APC).
In the current issue of this journal, Westerveld et al report the results of a systematic
review, aiming to report rates of complete eradication of intestinal metaplasia (CE-IM)
and dysplasia (CE-D) after CBA. The meta-analysis included seven clinical studies
with a total of 272 patients. The pooled rates for all outcomes were: feasibility
(95.8 %), CE-IM (85.8 %) and CE-D (93.8 %), defined as the proportion of patients
that achieved clearance of IM/dysplasia. The overall adverse event rate was 12.5 %.
Based on these results, the authors conclude that CBA is a safe and effective ablation
technique for treatment of BE neoplasia.
A systematic review with meta-analysis can give a balanced point estimate of all available
evidence, and is therefore considered the highest quality of evidence for a therapeutic
study [3]. However, a good systematic review should include studies that address the same
outcomes in comparable populations. For CBA, currently only a limited number of publications
are available, mainly feasibility studies. In our opinion, a systematic review and
meta-analysis at this time is therefore premature and inappropriate, as are the conclusions
drawn by Westerveld et al.
This meta-analysis is mostly a case of comparing apples with oranges, and a banana.
The banana in this case being the study by Canto et al, assessing outcomes of CBA
for squamous dysplasia. This study accounts for 120 of 272 cases in this meta-analysis
which aims to assess outcomes of CBA for eradication of IM and dysplasia [4]. Squamous dysplasia in the esophagus is a totally different disease entity than
dysplasia arising in BE. Inclusion of this study in this systematic review therefore
makes no sense at all.
When looking at the apples and oranges, the remaining six studies (152 patients) were
performed for eradication of BE with dysplasia. However, these studies answered different
questions and reported different outcome measures. Five studies report outcomes after
a single CBA treatment. An accurate outcome measure after a single treatment could
be the BE surface regression – an endoscopic assessment of the percentage of the treatment
area that converted to squamous epithelium at first follow-up. Three of these five
studies were reported by our group, and we indeed reported the BE surface regression
as primary outcome [5]
[6]
[7]. One should, however, not confuse this outcome of percentage of the BE segment eradicated
per patient after one CBA session with rates for CE-IM and/or CE-D (e. g. the proportion
of all patients that achieve CE-IM/D) after consecutive treatment sessions. Assessment
of CE-IM and CE-D only makes sense after consecutive treatment sessions with the aim
to eradicate the entire BE segment, yet these outcomes cannot be extracted from studies
in which only a single CBA treatment was performed as was the case in five of the
included studies.
Furthermore, it must be mentioned that most of the included studies were feasibility
studies. The results of these studies cannot be extrapolated to the general BE population
treated with ablative therapy. First, in most of the feasibility studies, only a relatively
small surface area was treated, with CBA applied either for a small BE island or short
segment BE (median C0–2 M0–5). Furthermore, in some studies, patients were included
after prior RFA treatment, not as refractory cases, but on the contrary as patients
with good regression and without complications after RFA. Logically, these patients
were likely to respond well to CBA as well. The results of these feasibility studies
therefore give an unrealistically positive outcome for efficacy and underestimation
of the complication risk. Whereas these feasibility studies are pivotal in the initial
phase of a new ablation tool, the results should not be extrapolated as benchmarks
for the technique in the general population.
If a systematic review and/or meta-analysis does not help us to answer what ablation
technique we should use for BE ablation, then what do we need? When a technique has
been shown to be feasible for conversion of BE to squamous mucosa in small series,
without technical issues or major complications, the next step is to assess sufficient
maturation of the technique. In our opinion this requires completion of at least one
large (i. e. >100 patients) single-arm, prospective, multicenter study that shows
that the technique is feasible and safe (i. e. serious complications in < 5 %). Such
a study should be accompanied with “endoscopic tips and tricks” learned during this
maturation phase. Once a new technique is evaluated in such a study, a logical next
step might be to compare the new ablation tool to the current gold standard of RFA.
Ideally, such a randomized study should be performed in centers with extensive experience
in treatment of BE neoplasia, among a homogeneous study population with dysplastic
BE, and after endoscopic resection of visible lesions, if present. After consecutive
treatment sessions, the outcome of interest should be the proportion of patients with
complete endoscopic eradication of BE and histologic confirmation by absence of IM.
Given the high point-estimate for RFA in expert centers for this outcome of >90 %,
a non-inferiority study would probably be the best fitting design. We recently performed
a power calculation for this imaginary trial and we showed that a total of 1,590 patients
should be included [8].
Should we undertake such an enormous investment in terms of time, money and patients,
which would give us, at best, an equally effective tool? Or would the outcomes of
such a study, which would run for many years, be already outdated by the time the
study has completed its follow-up? This study only appears useful if the new ablation
technique really holds promise for significant ancillary benefits over RFA. In our
opinion, it is questionable whether CBA or APC will provide such significant ancillary
benefits. One of the suggested benefits of CBA is improved tolerability, but one may
question whether this alone is sufficient. In none of the prospective RFA studies
post-procedural pain has been reported as a serious issue. Lower costs have been suggested
as a benefit for APC. However, this is purely based on the price of materials and
costs-efficacy analysis has never been systematically studied. Given the additional
time and treatment sessions required for eradication of a significant Barrett’s segment
with APC compared to RFA, the difference in costs may be less then generally assumed.
The authors mention that CBA might also be useful in patients that do not respond
to RFA, i. e. RFA-refractory cases. Indeed, instead of focusing on a new tool to replace
RFA, we might better explore whether these new techniques are complementary to our
current multimodality treatment approach. Although our current approach of ER and
RFA is very safe and very effective, we still have to deal with several challenges.
Improvements in the current treatment algorithm for BE with dysplasia and early cancer
could be found in new ablation tools that do work in RFA-refractory patients, that
can be used as rescue treatment in patients with persisting lesions that cannot be
removed with ER, or in treatment of patients with a stenosis that hampers passage
of the RFA catheter.
Conclusion
In conclusion, a new ablation tool first needs several small studies to prove that
it is feasible and these should be followed by at least one large, single-arm prospective
study to prove that the technique is mature. Before that, systematic reviews do not
provide reliable evidence for efficacy and safety of the technology.