The detection of disease progression remains a key issue in the management of well-differentiated small bowel NETs. In most centers, plasma CgA is used in conjunction with a variety of imaging. Although widely used, CgA exhibits significant limitations in terms of sensitivity and specificity and is not elevated in a substantial percentage (15-47%) of NETs [
3]. Imaging, both functional and topographical, is relatively insensitive in detecting alterations in indolent disease [
6] and histopathological analysis of resected specimens indicates that imagery fails to detect ~50% of lesions [
7]. Although the introduction of
68Ga-DOTA-PET and
64Cu-DOTATATE has amplified the ability to detect lesions, the former is not generally available and the latter is a research technique [
8]. Strategies for early detection of disease recurrence or progression that inform timely treatment initiation are therefore suboptimal [
9,
10].
Imaging and biomarkers
Imaging (CT, MRI, OctreoScan®,
68Ga-DOTA-PET/CT) are considered preeminent modalities to assess disease stability and progression of NELMs [
1]. There is, however, substantial variability in efficacy. The specificity for CT is as low as 22%, while both MRI and CT are negative in up to 50% of lesions [
11]. The sensitivity (69-86%) of
111In-octreotide scintigraphy is lower than
68Ga-DOTA-PET/CT (
68Ga-DOTATOC, -DOTANOC or -DOTATATE) [
12] which exhibits the highest sensitivity and specificity for NELM (82–100%; 67–100%) and extra-hepatic metastasis (85–96%; 67–90%) detection. In addition,
68Ga-DOTA-PET/CT detects lesions not identified by CT and/or MRI in up to 67% of patients [
6,
13].
18F-DOPA-PET and
11C-5-HTP-PET have some utility in functionally active NETs but are not publically available. Furthermore, they are not theranostics and do not possess a therapeutic counterpart [
14]. More recently, use of
64Cu-DOTATATE may surpass
111In and, theoretically,
68Ga in imaging sensitivity [
8]. Irrespective, it is apparent that >50% of all NELMs will be under-staged (pathological analysis of surgical specimens) [
7].
The use of individual peptides as biomarkers to identify early alteration in disease status has proved of limited value (e.g., pancreatic polypeptide) or amines (e.g., serotonin) although gastrin, glucagon and insulin are useful in specific NETs [
15,
16]. Overall, the most widely used is CgA which broadly correlates with hepatic tumor burden and survival [
17]. Elevations may be associated with tumor progression and in one report increased in 100% with progressive NELMs (disease relapse) [
18]. In a retrospective analysis, a reduction of ≥80% was predictive of complete resolution of symptoms and disease stabilization [
19]. In a separate study, CgA elevation was associated with residual disease [
20]. Problems with CgA include no relationship to tumor grade (which is prognostic for survival), concerns regarding sensitivity and specificity, and the absence of any universally accepted assay methodology [
3,
21]. The alternative, U5-HIAA, has limitations in terms of specificity and sensitivity [
22,
23]. Nevertheless, a reduction of U5-HIAA levels ≥80% (or normalization) is reported as predictive of symptomatic relief, but not of disease progression [
19].
Given the limitations of single agent biomarker analysis (CgA), we developed a multi-transcript (
n = 51 gene) molecular signature for PCR-blood analysis based on specific neuroendocrine tumor cell transcripts identified by mathematical analysis of 15 NET tissue microarrays [
4]. Gene co-expression network inferences and functional enrichment analyses of tumor tissue and peripheral blood NET transcriptomes (
n = 22) identified 51 candidate genes. A test set of NETs (
n = 130) was used to measure gene expression by hydrolysis-based qPCR and a tumor detection classifier was built using four learning algorithms (Support Vector Machine, Linear Discrimination Analysis, K-Nearest Neighbor and Naïve Bayes). This classification algorithm was validated in two independent NET sets (
n = 115,
n = 120) and exhibited a high sensitivity (85–98%), and specificity (93–97%) for NET detection including gastric, pancreatic and intestinal NETs. This significantly outperformed (ROC AUC: 0.95-0.98 vs. AUC: 0.64,
p < 0.0001) CgA measurements [
4]. Recently, this approach has been validated in a prospectively collected patient series [
24]. To quantify data we developed a classification algorithm - NET Index (0 = no disease,100 = active disease) [
25]. The index identifies progressive disease with a sensitivity and specificity of 91% respectively [
25]. In this case study we evaluated the utility of blood CgA levels (ELISA) and the peripheral blood hydrolysis-based qPCR of the 51 marker genes (NET Index) derived from in using imaging as a baseline comparator.
1) CgA levels
The first documented CgA measurement was made five years after initial diagnosis and was normal despite evidence of a mesenteric mass. Two years later, CgA levels remained normal despite a 0.5 cm NELM. CgA remained normal following cryoablation but became elevated after 2 months when bone and liver metastases were noted at PET-CT. Thereafter CgA levels normalized and remained within normal limits. Elevated CgA was only briefly detectable following cryotherapy when metastases were evident on imaging, but was normal when the hepatic metastatic burden was five lesions (>1 cm).
2) NET index
Circulating tumor transcripts were measured from the same samples (collected from 2008) as CgA. PCR analysis and establishment of the NET index product can be made within 8 hours of blood collection. The NET Index was elevated (95–100) from initial visit (December 2008) when residual tumor was evident by imaging (CgA was normal). After cryotherapy, CgA levels decreased (30%) but blood transcripts remained elevated and were elevated two months prior to imaging detection of additional metastases (April 2009). The NET Index remained high despite initiation of octreotide (20 mg, January 2009) and only trended down in May and November 2010 when PET-CT identified no disease to be present. Lower levels appeared to correlate with efficacy of octreotide-therapy. Transcript levels remained low until January 2011 when progressive increases in the NET Index were noted. The highest NET Index (November 2011) was also concordant with the elevated serotonin; at this time, CgA levels were normal. The NET Index elevations preceded the 68Ga-PET CT identification of five NELMs (February 2013). It should be noted that both a functional PET/CT with 11C-5-HTP (July 2011) and an MRI (January 2012) failed to detect disease at these time points. It is likely that the five lesions noted in (2013) were too small to be detected by PET/CT and MRI (July 2011, January 2012 scans).