| dc.description.abstracteng | Magnetic resonance imaging (MRI) in advanced rectal cancer (UICC stages II to IV) before and after neoadjuvant radiochemotherapy – its significance from the surgical point of view.
Hasibullah Habibi
Abstract
Background: In Germany treatment of advanced rectal cancer (clinically staged as UICC stages II to IV) consists of radical surgery (to achieve R0-status), preoperative radiation (RT) combined with chemotherapy (CTx) (to induce tumor remission) and postoperative CTx (to reduce metastatic spread). In several clinical studies this ap-proach has demonstrated improvement of survival accompanied by acceptable qual-ity of life.
For successful conduction of multimodality treatment comprehensive STAGING as well as RE-STAGING procedures are mandatory. These include rectal endosonog-raphy (rES, as preferred by colorectal surgeons) and pelvic magnetic resonance im-aging (MRI, as performed by specialized radiologists). But in daily clinical practice, the significance of MRI diagnostic is still under debate, especially from the sur-geons` point of view.
Purpose: In this thesis importance and accuracy of MRI (using a standard MRI 1.5 Tesla apparatus) at STAGING, RE-STAGING and for treatment monitoring were evaluated in three patients` cohorts (in total n=184 patients) with adenocarcinoma of the rectum (cUICC stages II to IV). All patients had received treatment at the Univer-sity Medical Center of the Georg August University of Goettingen (Lower Saxony, Germany) within interdisciplinary clinical trials.
Patients and methods: Between 04/2005 and 01/2010 36 patients of patients` co-hort A (control) received MRI at STAGING (MRI-1) followed by rectal cancer surgery performed as partial or total mesorectal excision (PME- or TME-surgery). In 60 pa-tients (patients` cohort B1) TME-surgery was conducted after completion of preoper-ative RT combined with 5-Fluorouracil (5-FU) CTx and in further 64 patients (pa-tients` cohort B2) after RT combined with 5-FU+Oxaliplatin CTx. In both patients` groups, B1 and B2, a RE-STAGING MRI-2 was performed preoperatively using the established imaging protocol of MRI-1. Furthermore, between 03/2011 and 08/2013 24 patients (patients` cohort C) were treated with intensified preoperative RT com-bined with 5-FU+Oxaliplatin and followed by 3 cycles of FOLFOX CTx. After comple-tion of RCT a RE-STAGING MRI-2 was conducted. After application of modified FOLFOX CTx a second RE-STAGING (MRI-3) followed preoperatively.
After PME-/TME-surgery including anterior (ARR), low anterior (LARR) and (extend-ed) abdomino-perineal resections (APR) pathological (p) confirmed tumor (T-), lymphnode (N- or LN-), residual (R-) and circumferential resection status (CRM-status) and other relevant clinico-pathological parameters were correlated with MRI-findings according to MERCURY-criteria and TNM-/UICC-classification system (8th and 10th edition). Furthermore, RCT+/-CTx-induced downcategorizing of UICC- and T-stages as well as the rate of pathological confirmed tumor regression grading (TRG, according to Dworak et al., 1997) were assessed. This TRG classification ranged from no regression (TRG 0 and TRG 1), intermediate regression (TRG 2 and TRG 3) to complete remission (TRG 4, equivalent to ypCR or ypT0).
At STAGING MRI findings (mr) were classified according to the TNM-/UICC-system (e.g. mrT-, mrN-, mrCRM-, mrTRG- status) and were compared to rES and several other clinico-pathological findings. Treatment associated shrinkage of tumor extent was measured by comparing the craniocaudal tumor length (in mm) in MRI-1 vs MRI-2 vs MRI-3. Furthermore, general tumor response to treatment was categorized using modified RECIST-criteria (complete (CR) and partial (PR) remission, stable (SD) or progressive (PD) disease).
In patients` cohort C, all results were correlated with disease-free survival (DFS), cancer-specific survival (CSS) and overall survival (OS) parameters after follow-up of 48 months (Median) using Kaplan-Meier-Logrank regression analyses. Further-more time-to-treatmet-failure (TTF) was determined.
For all cohorts MRI findings at STAGING (and RE-STAGING) were compared to postoperative histopathological findings as the reference gold standard. Binary STAGING (and RE-STAGING) parameters such as ≥ mrT3-status, ymrCR- (ymrT0), residual lymph node metastasis (LNM-) status as well as ymrCRM-status were demonstrated in four fields blackboards. Sensitivtity, specificity, the positive and negative predictive value as well as accuracy were calculated.
Results: In patients` cohort A (control) the mrT-status (MRI-1) was pathologically confirmed in 47%. The rate of overstaging was 37% and of understaging was 16%. The mrLN-status was correctly diagnosed in 42% of patients. In 53% an overstaging and in 5% an understaging was detectable. The mrCRM-status was correctly predict-ed in 92% at STAGING. In contrast, rES demonstrated a true prediction of uT-status in 60% of cases. The uT-status overstaging rate resulted in 30% and the uT- under-staging in 10%, respectively. For determination of the prognostic endpoint ≥T3-status, the sensitivity was 85% (MRI-1) and 91% (rES), the specificity in both 0% and the accuracy was 58% (MRT-1) and 63% (rES), respectively. Prediction of LNM in MRI-1 was possible with sensitivity of 86%, specificity of 17% and accuracy of 42%. For CRM positivity as preoperatively predicted by MRI, the sensitivity was 0%, the specificity 100%, the NPV 95% and the accuracy 95%, too. Finally, MRI-1 excluded early at STAGING any postoperatively detectable CRM positivity.
For correctly classified ≥ uT3-status at rES STAGING sensitivity amounted 91%, specificity 0%, PPV 69%, NPV 0% and the accuracy was 63%. In prediction of LNM (uN+ status in rES) the sensitivity was 70%, the specificity 45%, the NPV 75% and the accuracy 53%. In contrast to the excellent MRI findings according to the CRM, there was no adequate information with rES available.
In patients` cohort B the T-status and LN-status were correctly classified both by MRI-1 and rES in 65% and 58%, respectively. At STAGING with rES, rectal cancers were appraised uT2 tumors in 5%, uT3 tumors in 93% and uT4 tumors in 2%. MRI-1 evaluated mrT2 tumors in 7%, mrT3 tumors in 65% and mrT4 rectal cancer in 28% of cases. In contrast to rES with a detection rate of LNM in 75% of patients, STAGING MRI-1 allowed a much more precise subclassification of LN-status as shown in 14% mrN0-, in 66% mrN1- and in 20% mrN2-status of patients, respectively. A positive CRM (mrCRM+) was diagnosed in 55% of patients at STAGING in MRI-1. Preopera-tively, Re-STAGING MRI-2 (after RCT) correctly predicted ypT-status in 51% (52 of 101 patients). The ymrT-overstaging was 43% and ymrT-understaging was 6%. In MRI-2 ymrLN-status was correctly diagnosed in 60% of patients. The rate of ymrLN-overstaging resulted in 34%. Comparing MRI-1 with MRI-2, there was no change in the number of ymrT3 rectal cancers after RCT in patients` cohort B, but the amount of ymrT4 cancers decreased from 29% to 12%. Furthermore the percentage of ymrT2 tumors increased from 7% to 19%. In 5% of patients ymrT0-status (mrCR) was de-tectable. However, MRI-2 allowed prediction of RCT-induced tumor regression and verified significant reduction of craniocaudal tumor length compared to MRI-1 (p<0.001). Additionally, assessment of ymrLN-status predicted a RCT-induced re-gression in nearly 30% of cases and was confirmed as ypUICC-downstaging after pathological work-up of the resected rectal specimens.
In patients` cohort C, MRI-3 showed an ypUICC-downstaging in 67% of patients after intensified multimodality treatment. At least, an ypCR (ypTN0-status) rate of 17% was verified. In 20% of rectal cancer specimens no viable cancer cells could be de-tected (ypT0-status). Furthermore very encouraging survival data (data given in re-stricted mean survival time ± SE) could be achieved. At last day of observation post-operative DFS resulted in 49.3 ± 5.8 months, TTF in 57.1 ± 5.4 months and CSS in 62 ± 4.2 months, respectively. The OS was 54.7 ± 5.1 months.
After intensified RCT followed by 3 cycles of mFOLFOX-CTx a significant reduction in craniocaudal tumor extent was detectable comparing findings of MRI-1 vs MRI-3 (p<0.001). This treatment-induced tumor shrinkage was even predictable after com-pletion of RCT comparing tumor findings of MRI-1 to MRI-2 as well as between MRI-2 to MRI-3 (p<0.001 in each case).
Preoperative RE-STAGING MRI for prediction of ymrT0 status (ymrCR) showed a sensitivity of 31% (in cohort B) and of 0% (in cohort C), the NPV was 91% (in cohort B) and 79% (in cohort C), respectively. In both the specificity was 100% and all pa-tients without ypT0-status could be identified correctly. The PPV of 100% demon-strated all correctly predicted and pathologically confirmed CR (ypT0 status). In co-hort B the NPV as well as the accuracy for RE-STAGING MRI after RCT was 91%, respectively. Sensitivity in determination of CRM positivity or residual ypT4-status in preoperative RE-STAGING MRI was 100% in both, cohorts B and C. Accordingly the NPV was 95% (in cohort B) and 100% (in cohort C). All rectal cancer specimens preoperatively classified as CRM-negative could be confirmed histopathologically as cancer-free.
Discussion: Preoperative RE-STAGING MRI in detection of yT-status had a speci-ficity of 72% in patient cohort B and 73% in patient cohort C. These outcomes were below the results of a metanalysis recently published by de Jong et al. (2016). In his metaanalysis the rates of specifity ranged between 80% and 96%, as reported. This discrepancy might be explained by a higher rate of advanced rectal cancer in the presented cohorts B and C or by the use of more intensified multimodal treatment regimens. In the present study MRI findings at RESTAGING had an accuracy of 73% (in cohort B) and of 75% (in cohort C) in detection of yT-status and were only slightly below published data in the literature. For determination of residual LNM in RE-STAGING MRI the sensitivity was 75% (in cohort B) and 14% (in cohort C), re-spectively. The specificity was 53% (in cohort B) and 82% (in cohort C), respectively. The accuracy was 63%. However, these results remained below the results of the metaanalysis by de Jong et al. (2016) describing rates of accuracy for RE-STAGING MRI between 68% and 76% in their patients` cohorts.
In patients` cohort C of the present study, the specificity for ypN0-status was 82% at second RE-STAGING MRI. This important finding might indicate that the standard-ized use of MRIs after intensified multimodal treatment might allow a more precise classification and prediction of the relevant ypLN-status. This result should be re-evaluated in further clinical trials using intensified multimodality treatment regimens with higher patient case load.
Conclusion: In this monocentric study a standard MRI system (1.5 Tesla machine) with an optimized imaging protocol was used to clarify its significance in clinical rou-tine and from the surgical point of view. The MRI findings at STAGING resulted in better diagnosis and differentiation of advanced rectal cancers (≥T3 status), in a bet-ter (sub-) classification of LN-status, in measurement of tumor extent as well as in a more precise describing of cancer infiltration of anatomical surrounding tissue (e.g. mesorectal fascia system) or neighbouring pelvic organs compared to rES. Further-more MRI is the method of choice in RE-STAGING and MONITORING to evaluate efficacy of (intensified) preoperative multimodality treatment in rectal cancer. These results are in accordance with recently published data in the literature.
From the surgical point of view, routinely performed MRI is very helpful for early pre-operative determination of RCT-induced tumor shrinkage and of the extent of re-sponse to treatment (using mRECIST criteria). As a consequence of MRI findings the clinician will be able to identify non-responders very early and may change the previous treatment strategy consecutively. This might result in further intensification of treatment as well as in a complete attention to alternative therapeutic options (e.g. different CTx or sequence of drug application) or to a change of the initially planned radical TME-surgery to a more organ preserving intervention or a watchful procedure (e.g. in cases of excellent response to preoperative RCT +/- CTx).
Based on the results of this thesis pelvic MRI is the imaging method of choice for STAGING, RE-STAGING and MONITORING for patients suffering advanced rectal cancers and multimodality treatment. In ongoing clinical trials as well as in studies of the rectal-cancer-MRI-working-group at the UMG further parameters of prognostic relevance (e.g. mrTRG and extramural vessel infiltration, the so-called mrEMVI) are under evaluation and validation.
For further successful implementation of these clinical research projects a close in-terdisciplinary cooperation between surgeons, radiologist, pathologists, medical and surgical oncologists as well as general practitioners is mandatory. Additionally, an intensive collaboration with medical statisticians and biometrics is essential to eval-uate MRI findings continously in correlation with treatment results. Furthermore, all ongoing technical developments in MRI, resulting in large data packages, should be considered in these multidisciplinary teams. Consequently, the development of modern imaging procedures (e.g. real-time-funtional-MR-imaging) will early result in innovative treatment strategies and finally in best support of cancer patients. | de |