| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Correspondence |
Whilst we are pleased that Prof. Robinson has taken time to both read and comment on our paper [1], we feel that most of his comments are misdirected.
1. Inevitably new papers are published which may shed new light on a topic, but our review is still valid with respect to what was evaluated. Moreover, at the time of submission only one other relevant article (fulfilling selection criteria) had been published. We do not deny that technological advances in imaging continue apace, but the purpose of our systematic review was to establish the evidence for which is the best next imaging modality following ultrasound when a suspicious nodule is detected in a cirrhotic patient. This question is still relevant as most centres still use ultrasound (without contrast agents) as a first imaging technique.
2. As detailed in our methods we specifically set out to select articles reporting the characterization of liver lesions in cirrhotics, and thus those related to assessment of therapy, small studies (<10 patients), or when it was unclear if only cirrhotics were evaluated, were excluded. This excluded most studies but this does not mean that they are not a valuable contribution to the literature, just that they did not relate to the question we set out to evaluate.
3. We agree that the gold standard for reference to validate imaging for hepatocellular carcinoma (HCC), should be the explanted liver. We did indeed use this, but this was only evaluated in 10 studies (159 patients from 918) in the 29 studies selected; only 4 studies had direct comparison of imaging techniques. If Prof. Robinson had wished us solely to use this, then only 10 of 997 studies would have been selected. Thus we used other confirmatory criteria for diagnosis of HCC that are used in daily clinical practice: rising alpha fetoprotein or radiological evidence of tumour growth or diagnostic biopsy or confirmatory imaging (as defined by the authors themselves), as reference standards "to prove the lesion" was indeed HCC. These are not inadequate criteria as he implies.
4. The observer variation is an important issue and a relevant one. Prof. Robinson states that multiple observers improve diagnostic accuracy. This is true, but in practice a single radiologist reports a single image sequence for a patient, as he acknowledges. We did not exclude multiple observer studies, but did not come across a single study in cirrhotics (in our time frame) with confirmatory imaging or with the other criteria we have listed, in which multiple observers were used.
5. To make the paper readable, we felt we grouped studies in a sensible fashion and detailed the particular imaging techniques used. The very variation between studies, that Prof. Robinson points out, favours our conclusions in the discussion. Where the authors' results were corroborated either in relation to diagnostic impact or performance, we stated it, as these details were relevant ones for our study. The absence of such information was a limitation of the data available, as pointed out by us. The value of differential MRI contrast agents (tissue specific and extracellular) is a well known research interest of Prof. Robinson [2]. His paper with explant data after liver transplantation was published in March 2003 [3], after submission of our review. It may well be, that if this form of MRI is available, then it could be the next best test following an ultrasound. However, no comparison was made with other imaging techniques to prove it was superior.
6. We disagree about his comments on our conclusions. We couched them very carefully, suggesting ways in which reporting and evaluation of diagnostic imaging could be made in a uniform manner. He quotes our one sentence summary conclusion, by which we stand, which solely relates to the data we evaluated. However, we extensively preface this sentence with comments on assessing diagnostic impact and diagnostic performance as two separate issues, and provide examples of the difficulties encountered and the reasons why these exist. This makes it clear how the summary sentence should be interpreted and read. Furthermore, we suggest that studies assessing therapeutic impact might supersede diagnostic ones – a point that we feel Prof. Robinson should agree upon, as therapeutic modalities for HCC in cirrhosis depend heavily on accurate imaging diagnosis and staging.
7. We sense the frustration that Prof. Robinson has had, but respectfully point out that the basis for it is the poor reporting, poor evaluation and lack of published evidence concerning the subject matter and not the particular evaluation we have put forward. Thus any "post modern pseudoscience" resides in the data of most, but not all, the individual studies. Many unfortunately cannot be used conclusively to answer the question we posed. We accept that Prof. Robinson can have an opinion regarding our review as "tabloid journalism", but obviously disagree with his opinion as did the reviewers. We also strongly disagree with him that diagnostic methods cannot have outcomes as well defined as therapeutic techniques. This is with respect, patent nonsense: sensitivity, specificity and accuracy are key issues (as well as cost) for any diagnostic procedure [4]. Where the limits are set depends on the clinical context: the diagnosis of HCC in cirrhotics requires the highest possible optimum in all three parameters. This is particularly so, as therapeutic decisions are based on imaging, as often histology is cannot obtained (problems with clotting, anatomical position and seeding).
No doubt Prof. Robinson will want to disprove our conclusions in the future, concerning the current literature, by further publishing his own data. We hope he will succeed. If our review has spurred him on, then it has reached one of our objectives: better papers, better evaluation in papers, and eventually more streamlined and better diagnosis (needing less imaging) for cirrhotics with suspicious nodules on ultrasound examination. Lastly we feel that if Prof. Robinson re-reads our discussion carefully, (which can be an example of learning from experience, which he has emphasised to us), he will agree with what we say, as his own research aims in this area are designed to overcome the issues we raised.
Yours etc.,
1 Liver Transplantation and Hepatobiliary Medicine and 2 Department of Radiology, The Royal Free Hospital, Pond Street, London NW3 2QG
Received for publication December 15, 2004. Accepted for publication December 21, 2004.
References
Related articles in BJR:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
