modpathol2016248a PGL-PCC.pdf (6.38 MB)
Biologic and Genetic Basis of a Histological Risk Assessment of Early Recurrence/Metastasis in Paragangliomas
journal contribution
posted on 2017-03-23, 17:49 authored by Salvador J. Diaz-CanoSalvador J. Diaz-Cano, Nadia Talat, Alfredo Blanes, Klaus-Martin SchulteBackground:
Mid-term outcome information in risk stratified patient cohort is
needed to inform prognosis in individual patients with paragangliomas
(PGL), adjuvant therapy choice and future research. The objective is to
define the genetic features setting a PGL histological risk
stratification scheme of early recurrences/metastasis.
Design: A classification scheme for PGLs was devised, and specimens were
assessed for invasion capacity (infiltrative edges, extra-adrenal
extension, capsular and peritumoral vascular invasion), tumorigenic
expansion, necrosis, and mitogenic activity. Patients were prospectively
stratified as low histological risk (LHR) or high risk (HHR, presence
of at least one feature of invasive capacity and two features of
tumorigenic expansion). Patients underwent regular treatment and
follow-up for their PGLs in a tertiary referral center. Whole-exome
sequencing was performed on tumor and normal tissue samples, using a
Random Forest machine learning approach to compare LHR vs. HHR for early
recurrence/metastasis. Predictive Model Analysis Pipeline and Methods ►
Functional somatic mutations unique to tumors were identified and
represented as a samples x genes mutation matrix (mutated=1,
non-mutated=0). ► Pairwise Random Forest models were built for each
diagnostic category ► Variable selection was conducted using Fisher's
Exact test with 5x10 fold cross validation design, measuring predictive
accuracy in independent test sets.
Results: Disease-free survival was significantly lower in HHR patients
0% vs 78.4% (p=0.004). Histological risk stratification predicts DFS
with AUC of 0.8 (95% CI: 0.69-0.90; p<0.01). 7/37 HHR patients had a
synchronous diagnosis of malignancy based on other criteria and four
patients suffered a local recurrence. The HHR-predictive genes included
ARID1A (P=5.13E-09), AKT3 (P=3.57E-08), MAX gene associated
(P=6.86E-11), PIK3CA (P=1.35E-09), ELK3 (P=8.03E-16), MAP3K12
(P=4.17E-21), and TP53 (P=3.12E-40), being the accumulation of genetic
abnormalities the best predictor.
Conclusions: Histologic risk stratification of early
recurrence/metastasis is genetically defined by the accumulation of
alterations in the proliferation and cell survival pathways. A high-risk
status is associated with high risk of malignancy and disease
recurrence.