Muller Mirte, Hoogendoorn Roland, Moritz Ruben J G, van der Noort Vincent, Lanfermeijer Mirthe, Korse Catharina M, van den Broek Daan, Ten Hoeve Jelle J, Baas Paul, van Rossum Huub H, van den Heuvel Michel M
Department of Thoracic Oncology, The Netherlands Cancer Institute, Amsterdam, The Netherlands.
Department of Laboratory Medicine, The Netherlands Cancer Institute, Amsterdam, The Netherlands.
Tumour Biol. 2021;43(1):115-127. doi: 10.3233/TUB-211504.
The widespread introduction of immunotherapy in patients with advanced non-small cell lung cancer (NSCLC) has led to durable responses but still many patients fail and are treated beyond progression.
This study investigated whether readily available blood-based tumor biomarkers allow accurate detection of early non-responsiveness, allowing a timely switch of therapy and cost reduction.
In a prospective, observational study in patients with NSCLC treated with nivolumab or pembrolizumab, five serum tumor markers were measured at baseline and every other week. Six months disease control as determined by RECIST was used as a measure of clinical response. Patients with a disease control < 6 months were deemed non-responsive. For every separate tumor marker a criterion for predicting of non-response was developed. Each marker test was defined as positive (predictive of non-response) if the value of that tumor marker increased at least 50% from the value at baseline and above a marker dependent minimum value to be determined. Also, tests based on combination of multiple markers were designed. Specificity and sensitivity for predicting non-response was calculated and results were validated in an independent cohort. The target specificity of the test for detecting non-response was set at > 95%, in order to allow its safe use for treatment decisions.
A total of 376 patients (training cohort: 180, validation cohort: 196) were included in our analysis. Results for the specificity of the single marker tests in the validation set were CEA: 98·3% (95% CI: 90·9-100%), NSE: 96·5% (95% CI: 87·9-99·6%), SCC: 96·5% (95% CI: 88·1-99·6%), Cyfra21·1 : 91.8% (95% CI: 81·9-97·3%), and CA125 : 86·0% (95% CI: 74·2-93·7%). A test based on the combination of Cyfra21.1, CEA and NSE accurately predicted non-response in 32.3% (95% CI 22.6-43.1%) of patients 6 weeks after start of immunotherapy. Survival analysis showed a significant difference between predicted responders (Median PFS: 237 days (95% CI 184-289 days)) and non-responders (Median PFS: 58 days (95% CI 46-70 days)) (p < 0.001).
Serum tumor marker based tests can be used for accurate detection of non-response in NSCLC, thereby allowing early and safe discontinuation of immunotherapy in a significant subset of patients.
免疫疗法在晚期非小细胞肺癌（NSCLC）患者中的广泛应用已带来持久缓解，但仍有许多患者治疗失败并在疾病进展后继续接受治疗。
本研究调查了易于获取的血液肿瘤生物标志物是否能准确检测早期无反应情况，从而及时更换治疗方案并降低成本。
在一项对接受纳武单抗或派姆单抗治疗的NSCLC患者的前瞻性观察研究中，在基线及之后每隔一周测量五种血清肿瘤标志物。采用实体瘤疗效评价标准（RECIST）确定的六个月疾病控制情况作为临床反应的衡量指标。疾病控制时间<6个月的患者被视为无反应。针对每种单独的肿瘤标志物制定了预测无反应的标准。如果该肿瘤标志物的值较基线值至少增加50%且高于一个取决于标志物的待确定最小值，则将每个标志物检测定义为阳性（预测无反应）。此外，还设计了基于多种标志物组合的检测方法。计算预测无反应的特异性和敏感性，并在独立队列中验证结果。为了使其能安全用于治疗决策，将检测无反应的目标特异性设定为>95%。
我们的分析共纳入376例患者（训练队列：180例，验证队列：196例）。验证集中单一标志物检测的特异性结果为：癌胚抗原（CEA）：98.3%（95%置信区间：90.9 - 100%），神经元特异性烯醇化酶（NSE）：96.5%（95%置信区间：87.9 - 99.6%），鳞状细胞癌抗原（SCC）：96.5%（95%置信区间：88.1 - 99.6%），细胞角蛋白19片段（Cyfra21.1）：91.8%（95%置信区间：81.9 - 97.3%），以及癌抗原125（CA125）：86.0%（95%置信区间：74.2 - 93.7%）。基于Cyfra21.1、CEA和NSE组合的检测方法在免疫治疗开始6周后能准确预测32.3%（95%置信区间22.6 - 43.1%）的患者无反应。生存分析显示，预测有反应者（中位无进展生存期：237天（95%置信区间184 - 289天））和无反应者（中位无进展生存期：58天（95%置信区间46 - 70天））之间存在显著差异（p<0.001）。
基于血清肿瘤标志物的检测可用于准确检测NSCLC患者的无反应情况，从而使相当一部分患者能够早期且安全地停用免疫治疗。
