Surgery is a common and effective treatment for cancer of the colon. However, 10-20% of patients with this disease treated with surgery will go on to develop pulmonary metastases. In many cases, these can also be removed surgically, but the prognosis after resection of metastases is very variable and simple clinical parameters are unreliable as prognostic factors. A reliable tool for predicting the outcome of such resections would be very useful in aiding clinical decision making.

A nomogram is a two-dimensional diagram that can be used in calculation or in making numerical estimates. Several groups have developed, validated and used nomograms to estimate cancer patients’ survival probabilities from combinations of prognostic variables. These include a nomogram for estimating the probability of survival after resection of pulmonary metastases from colon cancer patients, first developed in the 1990s from a cohort of 313 patients treated at eleven different hospitals in Japan and validated using a slightly larger independent cohort of similar patients. It is known, however, that a prognostic test such as this one that is reproducibly valid with similar patients may not be transferable to an even slightly different environment. Yukihide Kanemitsu at the Aichi Cancer Center, Nagoya, Japan and his co-workers have now investigated the accuracy of this nomogram using a cohort of patients with metastatic colon cancer treated with advanced chemotherapy and surgery at a single centre [1].

The nomogram under assessment [2] can predict survival probability at one and three years and median survival time using primary histology; number of pulmonary tumours; lymph node enlargement; the existence of disease outside the thorax; and the level of carcino-embryonic Antigen (CEA) measured before pulmonary surgery as variables. Fifty-eight patients who had surgery at the Aichi Cancer Center between 1999 and 2005 were recruited onto the trial. The parameters above were determined for each patient during routine clinical examination and entered into the nomogram. These patients were found to differ from those from which the nomogram was derived in several significant ways, including a higher proportion diagnosed with moderately differentiated adenocarcinoma.

The nomogram was calibrated and then validated by comparing the predicted probability of survival for each patient with actual survival data. Patients were divided into four groups according to their nomogram-predicted risk. All patients were followed up for five years or until death and some for longer: 96.6% of patients were alive one year after pulmonary surgery, 70.5% after three years and 48.9% after five. Kaplan–Meier survival curves were calculated for each of the risk groups. Over all patients, a high nomogram score was strongly associated with poor survival. The Kaplan-Meier curves were clearly stratified, with the three- and five-year survival for each group within its predicted range. Overall, the nomogram had a concordance index (c-value) of 0.81, which indicates that the patient predicted to die first will do so in 81% of randomly selected patient pairs.

These results showed that a nomogram developed and validated using a large set of patients from a number of centres could predict the probability of survival equally well in a smaller and more recently treated independent cohort treated at the same cancer centre. Such a high predictive accuracy indicates that the tool may be more generally applicable as an aid to clinical decision making.

References

[1] Kanemitsu, Y.,Kato, T., Komori, K., Fukui, T. and Mitsudomi, T. (2010). Validation of a Nomogram for Predicting Overall Survival After Resection of Pulmonary Metastases from Colorectal Cancer at a Single Center. World J Surgery 34, 2973–2978. DOI 10.1007/s00268-010-0745-4

[2] Kanemitsu, Y., Kato, T., Hirai, T. and Yasui, K. (2004). Preoperative probability model for predicting overall survival after resection of pulmonary metastases from colorectal cancer. British J Surgery 91(1), 112-20.