Assessing the Predictive Value of Clinical Complete Response to Neoadjuvant Therapy for Rectal Cancer

Abstract & Commentary

Synopsis: In order to examine the accuracy of clinical evaluation, clinical and pathologic characteristics were compared for 488 rectal cancer patients who underwent preoperative chemoradiation. All patients were initially diagnosed with T3, T4, or node-positive disease based on physical exam or endorectal ultrasound. After neoadjuvant therapy and before resection, patients were re-examined clinically. Nineteen percent of patients had a complete response to neoadjuvant treatment based on preoperative clinical exam. Ten percent of patients had a complete response based on histologic exam of the resected specimen. Among the clinical complete responders, 25% had a pathologic complete response. Given its poor predictive value even in the hands of experienced clinicians, a clinical complete response must not preclude resection for rectal cancer patients.

Source: Hiotis SP, et al. J Am Coll Surgeons. 2002; 194(2):131-136.

The phenomenon of pathologic complete response (pCR) to neoadjuvant chemoradiation has led some investigators to question the dogma of surgical resection for nearly every solid tumor. Esophageal and rectal cancers are particularly amenable to preoperative chemoradiation and, thus, have been the focus of this controversy.1,2 The caveat is that pCR cannot be known without resection. Champions of chemoradiation only must rely on a clinical determination of complete response (clinical complete response; cCR). To address this issue, Hiotis and colleagues examined the positive and negative predictive value of cCR compared with pCR among rectal cancer patients treated with chemoradiation followed by resection.

Four hundred eighty-eight patients who received neoadjuvant therapy for stage II or III disease were included; patients were excluded if they had evidence of a second primary lesion. Patients had T3, T4, or node positive rectal cancer by endorectal ultrasound (ERUS) or fixed lesions by digital rectal exam (DRE) and proctoscopy. Treatment regimens included 2 cycles of 5-fluorouracil-based therapy and > 50 Gy radiation, followed by surgery in 4-6 weeks, consistent with the 1990 NIH consensus panel recommendations.3

Clinical response to neoadjuvant therapy was determined within 1 week of operation, which generally took place within 6 weeks of chemoradiation, although some patients waited as long as 12 weeks. A designation of cCR was given if no residual tumor could be identified by DRE or proctoscopy. Pathologic response was determined using standard methods to examine resected specimens. A designation of pCR was given to those specimens with bowel wall entirely absent of viable tumor cells. Chi square analysis was used to compare cCR and pCR rates.

The cCR rate was 19% (93/488); pCR rate was 10% (50/488). Among patients with cCR, 25% (23/93) had a pCR. That is, the positive predictive value of cCR = 25%. Among patients without cCR, 93% (368/395) also did not have pCR. That is, the negative predictive value of cCR = 93%. Notably, 4 patients with remaining node-positive disease on pathologic exam did not have residual tumor in bowel wall and were thus designated pCR.

Given the poor positive predictive value of cCR, even in the hands of experienced colorectal surgeons, Hiotis et al recommend that all stage II and III rectal cancer patients continue to undergo resection.

Comment by Arden Morris, MD

Focusing on rectal cancer, Hiotis et al address the recurrent question of when resection can be safely avoided. As in a host of previous studies, clinical evaluation proved inaccurate for assessing residual disease. However, PCR to neoadjuvant therapy is an encouraging phenomenon and deserves further careful study. Given the limited available evidence, the addition of postadjuvant preoperative ERUS, CT, or even MRI may not improve the positive predictive value of clinical assessment.4 Further study of these modalities and of PET scan in this setting is warranted.

Additionally, the implications of pCR are not completely understood. Long-term follow-up of pCR patients (who have, by definition, undergone resection) is still being gathered and recurrence rates, potentially from missed micrometastases, are unknown. Furthermore, if technology-enhanced determination of cCR can be closely correlated with pCR, future decisions to avoid resection would be based on outcomes of pCR patients who may have had residual unrecognized disease removed.

Still the impetus to eliminate surgical resection from the treatment of rectal cancer when possible is based upon worthy motives. Patients’ postoperative quality of life, risks of morbidity and mortality, and medical costs are important counters to the benefit of operation and must be acknowledged. For a few solid tumors, such as small cell carcinoma of the lung, squamous cell cancer of the anus, and some small squamous cell cancers of the head and neck, chemoradiation has been so successful that operation may often be virtually eliminated from the treatment regimen. For now, however, surgical resection remains the mainstay of treatment for rectal cancer and for most stage I-III solid tumors.

Dr. Morris is Robert Wood Johnson Clinical Scholar, University of Washington, Seattle, WA.


1. Coia LR, et al. J Clin Oncol. 2000;18:455-462.

2. Habr-Gama A, et al. Dis Colon Rectum. 1998;4: 1087-1096.

3. NIH. JAMA. 1990;264:1444-1450.

4. Kahn et al. Dis Colon Rectum. 1997;40:140-144.