A survival nomogram has been subsequently generated: it considers site, size, and MI as continuous variables, providing estimates stratified for patients aged ≤65 and >65 years.22 Bischof et al, in 2014, proposed a nomogram to predict disease-free survival (DFS) following surgical resection of GIST. They considered sex, tumor size, tumor site, and mitotic rate as prognostic factors. This nomogram stratified patients into prognostic groups and performed well on internal validation.23 In 2010, the first TNM classification for GIST was published. It was based on the classification of Miettinen et al,15 but it also considered the mitotic rate. There are eight stages corresponding to the eight subgroups of the AFIP classification (Figure 4).
However, ESMO guidelines do not recommend the use of this classification. Then, which is the most reliable and most easily used classification? Clinically, the classification score by Fletcher et al and Miettinen and Lasota20,15 are the most widely accepted. Several authors and the current ESMO guidelines have recommended the standardization of mitotic counting for GIST to avoid upgrading of patients’ risk. Schmieder et al24 reevaluated the impact of five widely applied and well established GIST risk classification systems (scores by Fletcher, Miettinen, Huang, Joensuu, and TNM classification) on a series of 558 GIST patients with long-term follow-up after R0 resection. This is a multicenter study encompassing 18 oncological centers in South Germany, from 2006 to 2012. Patients diagnosed before 2006 were registered retrospectively. After 2006, patients were recorded in a prospective manner. The results did not show any significant differences in terms of disease relapse prediction by all the scores, so they appeared to be equivalent. Anyway, the subgroup of high-risk patients was heterogeneous because of the different disease pathogenesis with consequences for tumor progression and clinical outcome.24 Heterogeneity in GISTs might not only be influenced by molecular, but also by other nongenetic factors (age, gender, tumor site, and syndromic occurrence), and maybe these factors should be considered in a prognostic score. Of course, standardization of mitotic count remains a basic principle for the score quality. Many authors consider the MSKCC prognostic nomogram more feasible in clinical practice than the other classifications. Chok et al tried to validate the MSKCC nomogram and to compare its predictive accuracy against other established risk classification systems, including the NIH, AFIP, and Joensuu criteria. They published a single-cohort retrospective study of 289 patients who underwent surgical resection for primary localized GISTs without adjuvant imatinib therapy and compared the actuarial RFS with the predicted RFS. The MSKCC nomogram and AFIP criteria had the best predictive accuracy for tumor recurrence compared with the NIH and Joensuu risk classification systems. However, the MSKCC nomogram slightly underestimated the probability of RFS after surgical resection of GISTs.17 Belfiori et al also compared MSKCC to NIH, NIH modified, and AFIP risk classifications. They confirmed the superiority of nomogram with respect to the other scores, even though it was not impeccable in predicting the RFS. Furthermore, they remarked that the main limitation of MSKCC nomogram remains the nonlinear consideration of mitotic count.25 Notably, the conventional clinicopathological parameters for risk assessment are poorly predictive of the outcome when dealing with “wild-type” GIST. In fact, SDH-inactivated GIST tends to have an indolent clinical behavior even in the advanced setting.26 Localized NF1-associated GISTs usually have a prolonged course but tend to become clinically aggressive once metastatic.27
Current guidelines recommend adjuvant therapy with imatinib in high-grade GISTs. Many classifications are useful to define risk stratification at the time of diagnosis, and it is very important to establish future treatment. High-risk GISTs are tumors that have a high number of mitosis, increase in size, and do not come from the stomach. Another important prognostic factor is tumor rupture. This is a very heterogeneous class of tumors, and for this reason several authors try to identify new prognostic factors. The biological field seems to be the most widely undertaken path in the literature. Radiological imaging could be another interesting field to analyze. Laparoscopic and open approaches are compared to verify different oncological outcomes. Even between small GISTs (tumors <2 cm in diameter), there are high-risk factors that should be carefully evaluated.
None of the many validated GISTs’ classification included any radiological findings as a prognostic factor, although there are some studies that have identified new risk factors. Zhou et al28 analyzed some computed tomography’s features of 129 patients with GIST >2 cm: primary tumor location, size, margin, shape, density, calcification, growth patterns, enhancement pattern, degree of enhancement, enlarged vessels feeding or draining the mass (EVFDM), necrosis, direct organ invasion, and lymphadenopathy. All these features were associated with the risk stratifications, as determined by univariate analysis. Only lesion size, growth pattern, and EVFDM remained independent risk factors in multinomial logistic regression analysis. The tumors with size >10 cm or 5–10 cm mixed growth pattern, or EVFDM were likely to be higher risk GISTs than those with size <5 cm, endoluminal growth pattern, or without EVFDM.28 Miyake et al assessed the potential value of preoperative 18F-FDG positron emission tomography to predict postoperative recurrence of solitary localized primary GIST after radical resection. Ring-shaped uptake and intense uptake were significantly associated with Joensuu high risk. Univariate analysis showed that ring-shaped uptake, intense uptake, size >5 cm, and Joensuu high risk were significantly associated with inferior RFS. Multivariate analysis showed that ring-shaped uptake (P=0.004) and Joensuu high risk (P=0.021) were independent adverse prognostic factors of postoperative recurrence.29 Another important tool for clinical evaluation of GIST is positron emission tomography/computed tomography. It facilitates both anatomic and functional evaluation of tumors and has become the standard imaging for GIST. Tokumoto et al determined the correlation of the risk category with standardized uptake values (SUV) max, tumor size, mitotic count, and MIB-1 index. The cutoff value for SUV max was found to be 3.0 between the low-risk and high-risk malignancy groups. Using univariate analysis in the high-risk malignancy group they determined that the SUV max value, mitotic count, and the MIB-1 index, but not tumor size, were predictive risk factors of malignancy. In a multivariate analysis SUV max was the only predictive risk factor for the high-risk malignancy group. This study recommended that submucosal gastric tumors with an SUV max >3.0 must be resected, even if the tumor size is <2 cm, because the tumors may have a high malignant potential.30
High-grade GIST group is a heterogeneous class. Therefore, many studies tried to identify biological markers as new prognostic factors. In particular, the aim was to identify prognostic markers that are able to select the best group of patients for adjuvant imatinib therapy. Many biological factors have been investigated. These factors are often related to high-grade GIST, but their prognostic value and their role to guide imatinib therapy are still debated (Table 1).
• The programmed cell death 1 (PD1)/programmed cell death ligand 1 (PDL1) pathway is a key inhibitor of the immune response. PDL1 expression was higher in AFIP low-risk samples than in high-risk samples. PDL1 expression was also higher in samples without metastatic relapse than in samples with metastatic relapse, suggesting favorable prognostic value.31
• Pfetin, a potassium channel protein, is a prognostic biomarker for GIST.32 Pfetin is an independent predictor of recurrence/metastasis for completely resected primary, localized GIST. There is an inverse relationship between Pfetin expression and risk of recurrence.33
• SETD2 may represent a novel GIST tumor suppressor gene, which contributes to GIST progression. SETD2 mutations are exclusively found in patients with high-risk/metastatic GISTs with a prevalence rate of approximately 11% but not in patients with low/intermediate risk. SETD2 mutations may occur at a later stage of tumorigenesis and appear to be associated with GIST progression, rather than initiation. Patients with GIST with SETD2 mutations or DNA methylation phenotypes showed shorter relapse-free survival.34
• SLITRK3 is one of the six isoforms of SLIT and neurotrophic tyrosine receptor kinase (NTRK)-like family member (Slitrk1-6), which are neuronal transmembrane proteins that control neurite growth. GISTs may originate from the interstitial cells of Cajal, with pacemaker potentials suggesting that mutations in genes involved in synapse or neural development may underlie GIST behavior. Increasing SLITRK3 expression correlates with decreased overall survival (OS) and DFS. SLITRK3 mRNA expression level increased according to NIH risk classification. SLITRK3 protein level is closely associated with tumor site, tumor size, and MI. Patients with high SLITRK3 expression, especially those who are also in the NIH high-risk groups, should receive IM adjuvant therapy and close follow-up management after surgery.35
• Dysregulation of miRNAs has been observed virtually in all major types of cancer, whereas the miRNA signature in GIST is not well characterized yet. miR-215-5 p is negatively correlated with the risk grade of GIST. The identification of miRNA profile GIST would be significant for further identification of new targets and development of novel therapeutics.36
• Monoglyceride lipase (MGLL) is a lipid metabolic enzyme causatively implicated in GIST progression. MGLL overexpression is associated with adverse clinicopathological factors and is independently predictive of unfavorable prognosis, suggesting its causative role in conferring aggressive phenotypes to primary, localized, imatinib-naïve GISTs.37
• Several lines of observations suggest that KIT/PDGFRA mutational status also impacts on GIST natural course, with most of PDGFRA-mutated GIST showing a more favorable outcome and GIST with a structural variant in the proximal region of KIT exon 11, for example, W557_K558del, or with exon 9 KIT mutation, that is, A502_Y503dup, behaving more aggressively.38,39