Background We developed a web-application, the so-called Sentinel, that allows early detection of cancer relapse or complications and early supportive care initiation/monitoring. It is based on a dynamical analysis of weekly self-reported patient symptoms which is used for triggering interventions. We conducted a phase III multicentric randomized study that compare the sentinel web-application to a routine imaging follow-up of advanced lung cancer patients.
Methods We randomly assigned patients with stage IIa to IV lung cancer who had no progression after their initial treatment to i) a routine imaging follow-up (CT-scan every 3 to 6 months) or ii) a sentinel web-application. In the experimental arm (Sentinel), data were weekly sent by using the web application between planned visits. An email alert was sent to the oncologist when self-scored symptoms matched some predefined criteria suggesting relapse or severe complication ; an imaging was then performed after a phone call for confirming the suspected symptoms. Early supportive care was also triggered and adapted according to the data transmitted by the patient. Overall survival, quality of life (FACT-L, TOI, PHQ9) and cost-effectiveness of both surveillance were assessed.
Results First planned intent to treat interim analysis of survival suggested tendency of a superiority of experimental arm. Per-protocol analysis suggested better survival in experimental arm. Ten-month survival was higher in experimental arm than in routine follow-up arm (85% and 70% respectively, p=0.04). Three-months quality of life tended to be better in sentinel arm versus standard arm (TOI mean score +3 vs -3 respectively, p=0,11). Performance status at the moment of relapse detection was 0-1 in 92% in sentinel arm versus 36% in standard arm (p=0.001). Mean time of web-application alerts management by oncologist was 2 minutes/10 patients using sentinel/week.
Conclusion In this provisional analysis of this phase III randomized study, survival and quality of life are better in experimental arm than in the routine imaging follow-up and without additional work for the physician. Other cancers may benefit from this personalized approach and further studied are in progress in lymphoma and solid tumors using specific relapse detection algorithms.