Abstract Rationale Assessing low-dose computed tomography (LDCT) lung cancer screening performance depends on correctly allocating diagnosed lung cancers as either LDCT “detected” or “interval” cancer. The principle criterion for this separation is whether the LDCT detected nodule(s) is suspicious for cancer. However, patients can develop interval cancers at different locations from their suspicious but non-cancerous nodules. In these cases, the LDCT would be incorrectly credited with detecting cancer. We conducted an analysis to determine how often such miscategorization might occur. Methods We analyzed data from 410 participants in the Watch the Spot Trial (NCT02623712), a cluster-randomized trial comparing surveillance strategies for pulmonary nodules (≤15mm). Participants were included if they had a newly identified nodule on an index diagnostic or screening chest CT scan and a lung cancer diagnosis within 27months of nodule identification. Location and laterality were collected from radiology reports. Concordance between index nodule and diagnosed cancer location was assessed using specific criteria: concordant (matching side and lobe), discordant (mismatched side or lobe), or ambiguous (side and lobe information agrees but one measurement contains an ambiguous categorization). We calculated overall concordance rates and evaluated location-specific concordance to determine what proportion of cancers in each lobe originated from index nodules identified in that same region. Results Participants were on average 68±9years old with balanced representation across genders (female:55%,225/410; male:45%,185/410). The majority were White (69.3%,225/410), the remaining were Black (14.9%,61/410), Asian (6.3%,26/410), and Hispanic (6.1%,25/410). Of the 402/410 cases with assessable data, 61.7% (253/410) cases were concordant between index nodule and cancer location, 29.3% (120/410) cases were discordant, and 7.1% (29/410) cases were ambiguous; 1.9% (8/410) with incomplete data were excluded from analysis. Location-specific concordance rates were: left lower 74% (35/47), right upper 69% (112/163), left upper 66% (63/95), right lower 66% (37/56), and right middle 40% (6/15). Conclusions We found 29% of diagnosed lung cancers appeared to arise from locations distinct from the identified index nodule, providing insight into the frequency interval cancers might be miscategorized as CT-detected. Limitations include: 1) location data were only collected for the largest nodule from participants with multiple nodules, thus some discordant cases may represent cancers arising from smaller nodules that were detected but not tracked, 2) these results only included subjects with a suspicious nodule that were subsequently diagnosed with lung cancer and therefore are not an estimate of interval cancer rates in LDCT screening populations. Further validation in larger prospective cohorts is warranted. This abstract is funded by: Patient Centered Outcomes Research Institute (PCS-1403012653)
Standley et al. (Fri,) studied this question.