- Design
- Preregistered prognostic model development and external validation using trial patient data
- Population
- 1,096 (CheckMate-214) and 651 (CheckMate-9ER) patients with untreated metastatic clear cell renal cell carcinoma
- Primary outcome
- Discrimination for 36-month overall survival
- Effect
- C-index 0.72 (95% CI 0.68 to 0.76) in validation vs 0.61 (0.57 to 0.65) for IMDC
IMDC risk groups were built in the targeted-therapy era but still guide first-line choice in metastatic clear cell renal cell carcinoma. This preregistered analysis used individual patient data from CheckMate-214 (1,096 patients) to build a new model, CPI2, and CheckMate-9ER (651) to validate it externally.
CPI2 uses 14 routine variables: age, Karnofsky performance status, previous nephrectomy, metastases to liver, lung, bone and lymph nodes, and seven blood tests (calcium, alkaline phosphatase, albumin, LDH, neutrophils, lymphocytes and white cell count). Its three risk groups discriminated 36-month survival with a C-index of 0.72 in both trials, against 0.65 and 0.61 for IMDC. Calibration was adequate. PD-L1 expression did not improve accuracy.
The model needs validation outside trials before it replaces IMDC in routine decisions, but it suggests IMDC classification is less reliable under checkpoint-based treatment than many assume.
- Continue to use IMDC for treatment decisions until CPI2 is validated in routine-care cohorts.
- Record all CPI2 variables at baseline; most are already part of standard work-up.
- Do not rely on PD-L1 expression for prognosis in metastatic clear cell renal cancer.
- Expect CPI2 to appear in trial stratification and reanalyses of checkpoint combinations.
Why it matters
The prognostic tool used to choose first-line therapy may be less accurate than assumed in the checkpoint era.
Don't overread it
Built and tested only in trial populations with good performance status; routine-care validation is still needed.
The statistics, in plain English
A C-index of 0.72 means the model correctly ranks which of two patients dies first about 72% of the time, against about 61% to 65% for IMDC here. Validation in a second, independent trial strengthens the result, though trial patients are fitter than routine populations.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for oncology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free