Bruce Sizer

Magnetic resonance-based texture parameters as potential imaging biomarkers for predicting long term survival in locally advanced rectal cancer treated by chemoradiotherapy

The study aimed to investigate whether textural features of rectal cancer on MRI can predict long‐term survival in patients treated with long‐course chemoradiotherapy.

Specialist surgeons and survival in breast cancer. More centralisation of services is not needed.

EDITOR,—Charles R Gillis and David J Hole may have found evidence that survival of patients with breast cancer is improved if they are treated by specialist breast surgeons, but there is little scientific evidence that surgery influences survival from breast cancer to any significant extent and they collected no information on the details of treatment received.1 There is considerable evidence, however, to support the survival …

The NHS in Clacton and the 2014 by-election

In The BMJ on Saturday 18 October, Ham wrote that “politicians are understandably reluctant to promise more than they can realistically deliver.”1 Contrastingly, on Sunday, Ed Miliband promised the public cancer diagnoses within seven days. On Monday, the Royal College of Radiologists’ press release said (I paraphrase) “blimey, you will need a lot more radiologists.” A weekend, let alone a week, is clearly …

Am I working in a “zombie hospital”?

Spiegelhalter quotes the Sunday Telegraph as reporting that “13,000 died needlessly at 14 worst NHS Trusts,” the 14 being those in the recent Keogh review, one of which was our trust.1 He notes how hard …

MRI for the assessment of locally advanced rectal cancer – a window of opportunity

Dear Sir, In recent years, there has been a paradigm shift in the management of rectal cancer and accurate imaging with MRI has become accepted as the modality of choice in defining preoperative disease extent [1]. However, in the management of patients with locally advanced rectal cancer [LARC: threatened ⁄ definite circumferential resection margin (CRM) involvement, T3 > 5 mm into perirectal fat, T4 tumours, and ⁄ or multiple nodes], there are concerns about the accuracy of preoperative MRI following long course chemoradiotherapy (CRT). We therefore read the paper with interest by Suppiah et al. [2], and would like to make several observations with regard to the methodology and data presented, and relate it to our own experience. We believe their conclusion that ‘MRI staging following CRT is poor’ may be flawed, because of the interval between the post-CRT MRI and definitive resectional surgery. In their paper, the median time from completion of CRT to MRI was 32 days i.e. 41⁄2 weeks, but surgery was performed at a median time of 55 days (8 weeks) i.e. just over 3 weeks later. A total of 86 patients were retrospectively analysed (as stated in their abstract), with complete data available on 64 patients; of these, 15 were subsequently excluded, including two patients who had inoperable disease despite being operable on MRI, and six patients (6 ⁄ 64, 9.4%) who were initially fit for surgery, but deteriorated either during CRT or between surgery and were deemed inoperable. Thus, only 49 of 86 patients (57%) underwent resection and are the subject of Suppiah et al.’s detailed analysis. We recently presented our 3 year experience on toxicity and response for 50 consecutive patients (36 men; median age 64 years, range 39–76 years), 47 with biopsy – proven LARC by MRI criteria and three patients with localized recurrence after previous anterior resection [3]. 48 patients (96%) completed CRT, of whom 43 (86%) underwent resection, of which 27 (63%) of which were performed laparoscopically. Thirtyseven patients (86% of those undergoing surgery) achieved an R0 resection. Importantly, only one patient developed liver metastases prior to surgery, with a median time to surgery from completion of CRT of 11 weeks (range 8–15 weeks). Of particular interest during this period, 17 patients with ‘bulky’ LARC, all of whom had CRM involvement on MRI, (extensively in six, all patients T3 or T4, ± multiple nodes), who received CRT and were felt to have an ongoing clinical response beyond MRI 2. Each had three high resolution MRI scans (before treatment, defined as MRI 1: 6–7 weeks after CRT, MRI 2; immediately preoperatively, MRI 3) [4]. All of these patients underwent definitive resection (15 anterior resection, 2 APR) 65–104 (mean 82) days post-CRT, with clear CRM in 16 patients (94%). MRI T and N staging and maximal wall thickness were correlated with the histopathological findings, with no patients having an increase in T or N stage. However, MRI 2 showed T downstaging in only 6% of patients, with a further 29.4% showing intra – T stage response from T3c to T3b. MRI 3 was performed 15–37 (mean 29) days after MRI 2, 3–21 (mean 10) days before surgery and showed further overall T stage response in 41%, and an another 17.6% had a further intra-T3 response: 50% of these responders had shown no T stage improvement on MRI 2. Applying an ordinal scale, T stage decreased from a mean of 3.57 at MRI 1– 3.3 at MRI 2– 2.51 at MRI 3 (P < 0.01; decrease in mean T stage between MRI 2 and 3, P < 0.01). In keeping with Suppiah et al.’s findings where T stage accuracy was 45%, with three times more patients overstaged on MRI rather than understaged, we found agreement between MRI 2 and final pathology was only 58.8%, with all seven inaccurately assigned patients overstaged. This is similar to other important UK studies recently, such as that by Allen et al. [5] where accuracy of prediction of T stage was 60%, N stage 70%, but median time from completion of CRT and 2nd MRI was 38 days, and median time from post-CRT MRI to surgery a further 43 days; similarly, Kulkarni et al. [6] also reported T and N stage accuracy at 43% and 78% respectively, but with MRI at 6 weeks (median), and surgery at 9 weeks. It is this gap between imaging and resection that may consistently undermine these studies, because in our series, MRI 3 T stage correlation was much higher at 82%. Fourteen of these 17 patients were also nodepositive by MRI criteria, with 50% response to CRT in seven patients by MRI 2, with no further N response on MRI 3 in the other 50%. Agreement with final pathology was 88%. Similarly, reduction in wall thickness was seen over the three scans (P < 0.01). In conclusion, we feel that preoperative MRI in the post-CRT setting is accurate if performed just before surgery, with a high correlation with final histopathology. Serial MRI can also document an ongoing response up to 12 weeks after CRT, and that waiting for this to occur appears not to be detrimental either from disease progression, or patients becoming unfit for surgery because of CRT. From both radiological and surgical