A nomogram to predict neutropenia in metastatic pancreatic cancer patients treated with gemcitabine/nab-paclitaxel

introduction: metastatic pancreatic cancer (mPC) is still associated with poor prognosis and limited therapeutic options. gemcitabine plus nab-paclitaxel (GnP) is a standard first-line chemotherapy associated with a 38% rate of grade 3-4 neutropenia. Moreover, grade 2 neutropenia is also frequent and demands dose-delaying. grade 2 to 4 (g2-4) neutropenia may thus significantly affect GnP efficacy. here we propose a nomogram to select patients (pts) at risk of g2-4 neutropenia who can benefit from the prophylactic use of granulocyte growth factors thus optimizing GnP efficacy. methods: a total of 74 full-dose GnP cycles (12 patients) were analysed. the following baseline data were collected: site of the primary tumor within the pancreas (head vs body/tail); the presence or absence of a biliary stent; presence of nodal, hepatic or pulmonary metastasis; body mass index (BMI); charlson comorbidity Index (CCI); count of neutrophils, lymphocytes and platelets; hemoglobin, creatinine, sGOT, sGPT, and total bilirubin levels. all these variables were assessed for grade ≥ 2 neutropenia prediction. neutropenia was graded according to the common toxicity criteria (CTC) v.4.02. results: median (range) pre-cycle neutrophils, platelets and hemoglobin were 4000/μL (1500-19000), 259000/μL (100000-717000) and 12 gr/dL (9.5-16.1), respectively. grade 2-4 neutropenia was recorded after 22% of cycles. in univariate logistic regression analysis (LRA) of 25 candidate predictors, platelets, neutrophils, and presence of lung metastasis were found to be significantly associated with the risk of g2-4 neutropenia (p-values from 0.07 to 0.02) and were assessed in a multivariable model. a multivariable LRA confirmed the three variables to be significantly associated with g2-4 neutropenia and were all included in a final predictive model (R2 index 23%, C-statistics 75%; P = .007). The model was internally validated with 100 boot-strap resamples (corrected R2 21%). the three variables were used to build up the nomogram with the following scoring system: absence of lung metastasis = 23 points; any 50000 decrease in platelets = 2.5 points starting from 0 points for ≥ 750000 platelets; any 2000 decrease in neutrophils = 2 points starting from 0 points for ≥ 20000 neutrophils. a score ≥ 126 was associated with a g2-4 neutropenia risk ≥ 25%. the 25% risk cut-off had the best discriminatory power according to a Receiver Operating Characteristic (ROC) analysis (sensitivity 81%, specificity 68%). A decision curve analysis revealed that for a threshold of risk of 25% or higher, there was an added net benefit of ≥ 37% patients adequately identified as at risk of grade 2-4 neutropenia by using the nomogram as compared to a ‘treat-all’ policy. In our cohort, patients with a predicted risk ≥ 25% had an incidence of grade 2-4 neutropenia of 41% as compared to 7% of patients with a predicted risk ≤ 25% (relative risk 5.69; P < .003) conclusion: mPC without lung metastasis and low pre-cycle neutrophil and platelet counts are at increased risk of GnP-induced grade 2-4 neutropenia. Patients with a predicted risk ≥ 25% according to the proposed nomogram should be considered for prophylactic use of growth factors.