Published online Jan 7, 2019. doi: 10.3748/wjg.v25.i1.118
Peer-review started: October 17, 2018
First decision: November 22, 2018
Revised: December 16, 2018
Accepted: December 19, 2018
Article in press: December 19, 2018
Published online: January 7, 2019
Processing time: 83 Days and 2 Hours
In recent decades, neoadjuvant therapy (NT) has been the standardized treatment for locally advanced rectal cancer (LARC). Approximately 8-35% of patients with LARC who received NT were reported to have achieved a complete pathological response (pCR). If the pathological response can be accurately predicted, these patients may not need surgery. In addition, no response after NT implies that the tumor is destructive, resistant to both chemotherapy and radiotherapy, and prone to having a high metastatic potential.
Few models or nomograms have been established and even fewer are used clinically to predict a good pathological response after NT for LARC. Therefore, developing accurate models to predict pathological response (PR) has great clinical significance and can help achieve individualized treatment in LARC patients.
Our goal was to establish nomograms that can be used to assist with individualized therapy as follows: for which patients NT and NT-related harm can be avoided; which patients will have good pathological responses to chemotherapy alone and radiotherapy can be avoided; which patients will have a good pathological response from a standard NT regimen, which patients need an enhanced mFOFOLX6-RT regimen; and which patients can use local resection or a “watch and wait” strategy to avoid complications. Solving these problems may aid in clinical treatment choices.
Our main objective was to establish nomograms for predicting a pathological response to different NT regimens based on pretreatment parameters for patients with LARC. We established accurate nomograms for predicting the pathological response to preoperative NT regimens based on pretreatment parameters for LARC patients. These nomograms can be used to distinguish patient types and facilitate developing individualized treatments.
Rectal cancer patients were identified from the database of The Sixth Affiliated Hospital, Sun Yat-sen University from January 2012 to December 2016. Four hundred and three patients who met the criteria were included. We collected all available clinical information before treatment.
The NT regimens included in our study were capecitabine/fluorouracil plus radiotherapy (standard group, capecitabine/deGramont-RT), mFOLFOX6 without radiotherapy (mFOLFOX6), and mFOLFOX6 plus radiotherapy (mFOLFOX6-RT). The radiation dose for the radiotherapy was 46.0-50.4 Gy, delivered as 1.8-2.0 Gy/d.
pCR was defined as no malignant cells found in the resected specimens, including the primary tumor and lymph nodes, and ypT0-2N0M0 (ypTNM 0-I) was classified as good downstaging.
Univariate logistic regression analysis was used to analyze variables related to the probability of pCR or good downstaging. Variables that achieved significance at P ≤ 0.05 in the univariate logistic regression analysis were further analyzed into the forward stepwise multivariable logistic regression. Multivariate logistic regression analysis was used to construct the nomograms. Because the NT regimen was a statistically significant factor for predicting pCR probability, we then attempted to develop three nomograms based on the different NT regimens to predict pCR probability. The C-index was acquired for the nomogram, and internally validated using the bootstrap method to determine the adjusted C-index. Calibration curves of the nomograms were generated to show the relationship between the predicted and observed outcomes.
All statistical analyses were performed using SPSS 24.0 and R 3.5.1.
Of the 403 patients in our study, 281 (69%) were men. As assessed pathologically, 72 (17.86%) individuals achieved a pCR; 177 (43.9%) patients achieved ypTNM 0-I and were classified as having good downstaging.
Significant differences were found for age, tumor differentiation, TL, DTAV, mesorectal fascia (MRF) status, interval, and NT regimen in the univariate analysis. In the multivariate analysis, NT regimen types, tumor differentiation, TL, and MRF status were significantly associated with pCR probability.
Significant differences were found for carcinoembryonic antigen (CEA), tumor differentiation, distance of tumor from the anal verge (DTAV), tumor length (TL), cT, and MRF status in the univariate logistic regression analysis for good downstaging. In the multivariate analysis, tumor differentiation, MRF statuses, and cT were significantly associated with the probability of good downstaging.
Table 5 shows the distribution of pretreatment clinical parameters in the NT regimen groups. No differences were found in any factors between the three groups except age and DTAV.
In the univariate analysis of the capecitabine/deGramont-RT group, NLR was the only significant factor for predicting pCR probability. NLR (> 3) was the only significant factor compared with NLR ≤ 3 in the further multivariate analysis. We could not develop a nomogram to predict pCR probability in this case.
In the univariate analysis of the mFOLFOX6-RT regimen, TL and MRF status were significant factors predicting pCR probability. TL and MRF(+) were significant factors in multivariate analysis.
In the univariate analysis of the mFOLFOX6 regimen, tumor differentiation and TL were significant factors for predicting pCR probability. Further multivariate analysis showed that differentiation and TL were significant factors.
Nomograms were developed based on the significant factors in the multivariate logistic regression analysis. We used 1000 bootstrap resamples to compute an adjusted C-index. Calibration curves between predicted and actual observations by internal validation demonstrated that these nomograms showed good statistical performance for predicting the probability of pCR and good downstaging.
We established accurate nomograms to predicting the pathological responses to different preoperative NT regimens based on pretreatment parameters for LARC patients. These nomograms can be used to distinguish patient types and facilitate developing individualized treatments.
To the best of our knowledge, our study is the first to use different NT regimen types to predict a pathological response. We established an accurate model with easily obtained variables to predict the probability of pCR and good downstaging. Our analysis was also strengthened through cross-validation. These models can be used to assist with individualized therapy as follows. For LARC patients expected to have a poor pathological response, NT and NT-related harm can be avoided. For patients expected to have good pathological responses to chemotherapy alone, radiotherapy can be avoided. For patients who are not expected to have good pathological response from a standard NT regimen, an enhanced mFOFOLX6-RT regimen can be considered. For patients with a high probability of pCR after NT, local resection or a “watch and wait” strategy can be used to avoid complications.
Our analysis had several limitations. First, this was a retrospective study, in which some factors associated with pCR were unavailable, such as smoking status, molecular subtypes and so on. Second, mFOLFOX6 and mFOLFOX6-RT are not the standard regimens for LARC, and both regimens remain in the clinical trial phase. Finally, our nomograms are based on the experience of our single institution. These results must be validated in a group of independent external institutions.
The nomograms established in our study can be used to evaluate the probability of a pathological responses before NT and after NT. However, additional studies are required to answer clinical questions regarding which patients can be treated only with neoadjuvant chemotherapy, which patients need oxaliplatin added to the neoadjuvant chemoradiotherapy, which patients need radical surgery, which patients can undergo local excision, and which patients can be managed with a “watch and wait” strategy after achieving a good response.
In the future, we plan to include a larger number of patients to enhance the accuracy of the prediction. On the other hand, we plan to add a second external cohort for validation to strengthen the reliability of the nomogram.