Patterns of recurrence in patients achieving pathologic complete response after neoadjuvant chemoradiotherapy for rectal cancer

Following the introduction of TME and CRT, local failure rates after curative (R0) resection of locally advanced rectal cancer have decreased dramatically (Sauer et al. 2004; Collette et al. 2007; Wibe et al. 2002). Previous studies have consistently demonstrated that patients with good response after CRT have better long-term outcomes (Garcia-Aguilar et al. 2003; Park et al. 2012).

The AJCC and International Union Against Cancer recommend retrieving a minimum 12 LN to achieve appropriate staging. However, several studies have reported lower LN yield in proctectomy specimens from patients treated with preoperative CRT (Fuente et al. 2009; Marks et al. 2010; Govindarajan et al. 2011). In the present study, we retrieved much fewer LN (median number of examined LN: 8, range: 0–18) than previous reports of patients treated with neoadjuvant CRT. As has been previously reported, the number of LN harvested varies greatly even when surgical treatment and pathological evaluation have been standardized (Marks et al. 2010). At our hospitals, all colorectal surgeons involved in this study had received formal training in TME, thus all patients in this group underwent surgical treatment involving adequate TME proctectomy. The inadequate LN sampling may suggest incomplete TME, which might result in adverse oncological outcome. However, some researchers were unable to identify any relationship between the reduced numbers of LNs examined and decreased patient survival (Rullier et al. 2008; Ha et al. 2010). Moreover, a study by the Cleveland Clinic not only confirmed the association between neoadjuvant therapy and decreased number of LNs retrieved from proctectomy specimens, but also suggested that the diminished number of LNs is actually an indicator of improved tumor response and consequently decreased local recurrence rates (Campos-Lobato et al. 2013).

LPLN metastasis is another reason for the local recurrence in our data. In the present study, the failure pattern of the three cases raises the question of whether LPLD should be performed in patients with suspected LPLN metastasis based on pre-CRT CT or MRI, and how patients who may benefit from LPLD can be identified. Studies from Japan have demonstrated that the incidence of LPLN metastasis is approximately 15% in patients with advanced low rectal cancer who underwent LPLD, and varies from 10.6 to 25.5% (Konishi et al. 2010). The major drawback is that the above mentioned results were obtained from patients who had not undergone preoperative CRT. In the Western countries, the standard treatment for patients with locally advanced rectal cancer is preoperative CRT followed by TME alone. LPLN metastasis is generally considered a metastatic disease that is not amenable to surgical cure. Moreover, some studies have revealed that preoperative radiotherapy followed by TME led to acceptable local control that was as good as that achieved by extended surgery, which included LPLD (Georgiou et al. 2009; Quadros et al. 2012). Consequently, it is widely accepted in the Western countries that routine LPLD is not necessary. However, several studies have revealed that selective LPLD might improve local control and survival in patients with LPLN metastasis in advanced low rectal cancer who have been treated with preoperative CRT. Akiyoshi and colleagues showed that the rate of pathological LPLN metastasis was 66% when they performed pre-CRT LPLD based on LPLN size and that preoperative CRT alone could not eradicate LPLN metastasis completely, thus LPLD might improve local control and survival of patients with LPLN metastasis in advanced low rectal cancer treated with preoperative CRT (Akiyoshi et al. 2014). Kim and colleagues showed that lateral pelvic recurrence is the most common pattern of local recurrence following neoadjuvant CRT plus TME alone and that patients with preoperative evidence of LPLN metastasis could benefit from LPLD even after neoadjuvant CRT (Kim et al. 2008). In our study, among the 195 patients who achieved pCR, CT scan or MRI detected lateral LN metastasis in ten patients with low rectal cancer, 3 (30%) of whom eventually developed lateral pelvic recurrence after follow-up. Although the evidence from our cohort is very limited, we agree with previous studies on the most common pattern of local recurrence, and that TME alone following preoperative CRT might be insufficient for controlling LPLN metastasis that is present before CRT. As suspicious metastatic LNs in the lateral pelvic area are a risk factor for local recurrence in patients who have undergone curative resection with preoperative concurrent CRT, and LPLD has been associated with increased morbidity such as blood loss, and urinary and sexual dysfunction, correct detection of LPLN is crucial. Previously, CT was the standard method for assessing the pre-treatment clinical nodal stage. Unfortunately, a meta-analysis (Bipat et al. 2004) reported only 55% sensitivity for LN involvement by CT staging. Consequently, more reliable imaging criteria for detecting LPLN metastasis are needed. Currently, newer techniques, e.g., MRI or positron emission tomography (PET) scanning are used to improve LN staging accuracy. Brown et al. (2003) found that high-spatial resolution MRI had a sensitivity of 85% [51/60 (95% confidence interval, CI: 74%, 92%)] and a specificity of 97% [216/221 (95% CI: 95%, 99%)] for LN (+) defined by the presence of mixed signal intensity or irregular nodal capsule border. Koh and colleagues (Koh et al. 2010) used ultra-small paramagnetic iron oxide (USPIO), a LN-specific biodegradable contrast agent with the potential to enable identification of nodal metastasis independently of LN size for MRI. They found that the USPIO enhancement pattern had higher diagnostic specificity, but the same sensitivity, as morphological findings in pathologically matched mesorectal LNs (specificity, 93 vs. 75%; sensitivity, 65 vs. 65%). It appears that PET/CT is less accurate than MRI for LN staging (Farwell et al. 2014), but has higher specificity than MRI (83–85% vs. 67%) in nodal staging (Kim et al. 2011; Cipe et al. 2013). Additionally, we believe that combined MRI and PET (fusion imaging) would increase the accuracy of nodal staging. Further studies are essential for identifying the most reliable imaging criteria for LPLN metastasis in patients treated with preoperative CRT.

The present study demonstrates that distant metastases are the major failure pattern of patients who achieve pCR after CRT, which is similar to the findings of previous reports (Table 2). Maas and colleagues reported a 5-year distant metastasis rate of 11.2% (Maas et al. 2010). Capirci and colleagues performed a long-term analysis of 566 patients who achieved pCR, reporting a distant metastasis rate of 8.9% after a 46-month follow-up (Capirci et al. 2008). The persistence of distant metastases suggests the inadequate systemic control of the CRT regimen. Thus, the high rate of distant failure may indicate that adjuvant chemotherapy is warranted in these patients. However, the role of postoperative CT following neoadjuvant CRT and radical surgery for patients with locally advanced rectal cancer is still not entirely clear, especially in patients with pCR. Some studies suggest that there is little or no benefit from chemotherapy (Maas et al. 2010; Capirci et al. 2008; Kiran et al. 2012). However, Collette and colleagues (Collette et al. 2007) demonstrated that the administration of adjuvant chemotherapy was associated with significantly improved overall survival. Their study randomly assigned patients to adjuvant chemotherapy or no adjuvant chemotherapy after (chemo) radiation and surgery for rectal cancer. Patients with ypT0–2 tumors, but not ypT3–4 tumors, benefited from adjuvant chemotherapy; the authors explained that tumors that respond well to preoperative FU-based chemoradiation would also be sensitive to adjuvant chemotherapy involving this drug. These contrasting findings indicate that more trials involving adjuvant chemotherapy following neoadjuvant CRT should be performed in future.

Consistent with previous reports, we also observed a unique pattern of distant recurrence, where there was a predominance of non-liver metastases. In our study, only one patient developed liver metastasis; extra-liver metastases involved seven lung metastases and two bone metastases. Different theories have been proposed to explain the pattern of metastases, although this process appears to be multifactorial (Ding et al. 2012). The “seed and soil” hypothesis states that, due to different organ-specific microenvironments, tumors may metastasize to specific organs independently from the vascular anatomy (Fidler 2003). However, some researchers suggest that tumors might metastasize to specific organs independently based on the differential level of thymidylate synthase (TS) mRNA, which causes differential sensitivity to 5-FU-based chemotherapy. Patients with higher TS mRNA levels easily develop pulmonary metastasis (Yamada et al. 2001). Such findings are also attributed to the nature of the venous drainage of the rectum (Watanabe et al. 2011). Intensive surveillance schemes and improved imaging technology may also contribute to the high rate of lung metastasis detected, which is consistent with that of previous studies (Kirke et al. 2007; Lee et al. 2007).

There are some potential limitations to this study. The first lies in the limited number of enrolled patients, given their good outcomes. The second potential limitation is that the number of LNs retrieved was much lower than that in previous reports on patients treated with neoadjuvant CRT. However, the inadequate LN sampling in our study may have been due to improved tumor response. The third potential limitation involves regular assessment of the circumferential resection margin (CRM) status. The CRM is considered one of the strongest predictors of surgical failure in rectal cancer (Kelly et al. 2011). We do not routinely assess the CRM status, but the pathologists at our hospitals routinely perform macroscopic evaluation of rectal cancer resection specimens. The quality of the mesorectum is classified as good (mesorectal), intermediate (intramesorectal), or poor (muscularis propria plane) (Quirke 2003). The mesorectum quality of the 18 patients included in our data was good, indicating smooth CRM on slicing.

In conclusion, our data demonstrate that distant metastases are the major failure pattern of patients who achieve pCR after CRT and that non-liver metastases are the predominant pattern of distant recurrence. Our results indicate the possible predominant pattern of distant recurrence and serve as a basis for managing LPLN metastasis in low rectal cancer. Patients with LPLN metastasis in advanced low rectal cancer treated with preoperative CRT would benefit from selective lateral LN excision.