Survival Calculator Takes ‘Holistic’ Approach in Oral Cancer

MONTREAL – A predictive calculator aims to offer patients with oral cancer a personalized estimate of survival that takes into consideration the effects of co-existing conditions.

In a presentation at the American Head & Neck Society (AHNS) annual meeting, Louise Davies, MD, of the VA Outcomes Group in White River Junction, Vermont, gave the example of patients, around age 70, who were diagnosed with oral squamous cell carcinoma (OSCC).

“In this case, 5 years after diagnosis for our patients, 36 out of 100 would be alive, 41 out of 100 would die of their oral cancer, and 23 out of that 100 patients would have died of something else,” she said. Details on the calculator were simultaneously published in JAMA Otolaryngology-Head & Neck Surgery.

“The models in the calculator showed that patients with oral cavity cancer had a higher than average risk of death from other causes than the matched US population, and this risk increases by stage,” wrote Davies and colleagues.

They noted that the “Surveillance, Epidemiology, and End Results (SEER) Program Oral Cancer Survival Calculator (SEER OCSC) supports a holistic approach to the life of the patient, and the risk of death of other causes is treated equally to consideration of the probability of death of the cancer.” The calculator, which is intended for use by people with oral cancer and their clinicians, is available online.

Davies said the calculator applies to those ages 20-86 who are newly diagnosed with OSCC, excluding tonsil or tongue base cancers and cancers associated with human papilloma virus infection.

Her group also set out several caveats with the calculator, stressing that the “calculations and estimates provided by the SEER OCSC represent average individual survival; they do not estimate what the outcome might be for that individual person.” Also, “the SEER OCSC should not be used for guiding decisions about specific cancer treatments.”

And limitations of the multi-model-based study done to develop the calculator included the fact that cancer staging data were from the American Joint Committee on Cancer (AJCC) 6 versus the updated AJCC 8 “which just came into use in 2018…There is not yet sufficient follow-up time available to use AJCC 8,” the researchers said. Also, “there are variables that we have chosen not to include, such as pack-years of smoking and alcohol intake,” both of which are tied to a higher risk for oral cancer.

‘Still Opaque’

In an accompanying invited commentary, Carole Fakhry, MD, MPH, of Johns Hopkins School of Medicine in Baltimore, and colleagues reiterated that “the survival calculator is not intended to guide discussion surrounding specific cancer treatments. Non-randomization of treatment assignments that may be associated with a patient’s health status and the specific geographic and practice patterns through which patients are offered treatments are major barriers to incorporating the differential effect of competing treatments into this calculator.”

They pointed out that predictive oncologic survival models must balance “[a]chieving accuracy by including numerous factors and complex mathematical formulas versus succeeding in simplicity, usability, and communication, perhaps at the sake of accuracy. All estimation techniques are fraught to some degree by bias, uncertainty, and variability.”

Fakhry’s group advised clinicians to take the predictive information from SEER OCSC and share it with patients “with grace and sensitivity. Potential risks of revealing personalized prognostic survival estimates to patients include increased anxiety and distress surrounding competing causes of death, misinterpretation of data (particularly among those with limited health literacy), loss of hope and engagement in cancer treatments, and overemphasis on statistics that may overshadow individual preferences and circumstances.”

They emphasized that tumor response to treatment cannot be determined in advance, and that “underpins the challenges in prognostication models. With immune-oncology treatment paradigms, response patterns can differ markedly from traditional anticancer treatments. We do not yet have enough collective experience and longitudinal data to understand which patients may experience durable immunotherapy responses, flipping the switch from potentially lethal to chronic disease.”

Finally, clinicians and patients need to consider that quality of life is “an important treatment outcome that has yet to be captured in current prediction calculators. While we all wish we had that perfectly clear crystal ball, even the best prediction tools may still be opaque,” Fakhry and colleagues said.

Study Details

Davies and colleagues used data from the SEER 18 registry (2000-2011), SEER-Medicare linked files, and the National Health Interview Survey (NHIS, 1986-2009).

“Statistical methods developed to calculate natural life expectancy in the absence of the cancer, cancer-specific survival, and other-cause survival were applied to oral cancer data and internally validated with 10-fold cross-validation,” they explained. The main outcome and measure were the probabilities of surviving or dying from the cancer or from other causes, along with life expectancy if a person did not have cancer.

The researchers reported that they included 22,392 patients with OSCC (60.5% male; 78.1% white) and 402,626 NHIS interviewees. The mean age at diagnosis for people with stage I cancer was 61.6, while for stage II, it was 64.7, and 62.9 and 62.3 for stages III and IV, respectively. The most comorbidities were chronic obstructive pulmonary disease and diabetes.

Davies’ group offered another example of how the calculator worked, explaining that “conditional on having survived to age 50 years, a female and male patient diagnosed with stage III cancer would have a 60% and 44% chance, respectively, of being alive at age 70 years, in the absence of their cancer. In the general US population, the corresponding estimates are 86% and 79%, respectively, an absolute difference of 26 and 35 percentage points, respectively.”

In an accompanying special communication, Davies and colleagues acknowledged that a “concern about this calculator might be whether the age of the underlying data affects the accuracy of the estimates, especially if risk factors, treatment patterns, diagnostic precision, or care for other conditions have changed,” but argued that their comparison of earlier SEER data showed that “estimates obtained from the calculator are not likely to be substantially different from what current trends would indicate.”

They pointed out that there are other prognostic calculators available in oral cancer that “may provide complementary information to this tool.”