Introduction

GISTs, at one time, were thought to be quite rare, but because of an increased ability to reliably diagnose them, their incidence is now estimated at around 5000 new cases per year in the USA, which place them among the most common sarcomas [1].

It is considered that most GISTs have the potential to recur and grow in a diffuse way even after their complete excision and to acquire the ability of distant metastasis. In 2002, the National Cancer Institute (NIH) established a consensus for evaluation of their recurrence risk [2]. This scheme includes the anatomical site where the tumor is presented, the major diameter of the tumor and the number of mitosis in 50 high-power fields (50 HPF); however, a modification of this scheme obviated the location of the tumor. Nevertheless, there are cases in which one or more of these data are unknown, making it impossible to predict their risk of progression. It is therefore necessary to find different prognostic factors or a risk stratification system to apply them, by example, to cases where 50 HPF cannot be counted. Moreover, pre-operative assessment of GIST malignancy is not easy, and hence a more practical approach would be valuable.

The recent development of imaging studies and the simplicity of determining tumor size by volumetry could be useful and practical for evaluating patients with GISTs, because the size in the risk assessment of GISTs refers to the single largest dimension, but GISTs are usually (almost every case) irregular tumors, and using a single dimension is an oversimplification of the more complex tumor. We hypothesize that a risk stratification system based only on tumor volumetry (measured with computer tomography and/or in the surgical specimen) could predict the risk of recurrence in patients with GIST, better than the NIH consensus criteria based on size and mitotic count. The research question was whether the tumor size determined by volumetry allows a better risk stratification than the traditional system, and our aim was to compare the diagnostic performance for recurrence of NIH consensus criteria against risk stratification based on tumor volume in uninodular, previously untreated primary GISTs.

Material and Methods

This work was approved by the Research Ethics Committee of the National Cancer Institute of Mexico (approval number: Rev/70/17). Surgical specimens of primary (non-metastatic) uninodular GISTs without prior treatment were selected from the pathological files of our Institution between 1995 and 2015 relating to patients older than 18 years, who had at least 1 year of clinical follow-up, with no prior treatment (including imatinib) (n = 74). Clinical and histological characteristics were recorded from the patients’ clinical files.

The patients were classified into four risk groups, following the NIH consensus criteria (Table 1) based on the tumor size in largest dimension and mitotic count in 50 HPF. With the three largest diameters of the tumors measured in the surgical specimens and in the pre-surgical tomography (taking the diameters settled in the radiology report), tumor volume was calculated with the formula of the volume of an ellipsoid ((4/3) × π × r1 × r2 × r3). Then, a ROC curve was created to identify cut-off points of tumor volume determined in the surgical specimen associated with recurrence. Based on the ROC results, tumors were grouped into three categories, both for the surgical specimen (0–250 cm3, 251–1200 cm3, and > 1200 cm3) and for the pre-surgical tomography image (0–200 cm3, 201–1150 cm3, and > 1150 cm3).

Table 1 National Institute of Health (NIH) Consensus Criteria for recurrence risk stratification of Gastrointestinal Stromal Tumors
Full size table

Basal features of the patients with recurrence were compared with those of patients without recurrence during follow-up. For the comparison of numerical variables and based on a normality test (Kolmogorov-Smirnov test), a Mann-Whitney U test was performed. For the comparison of qualitative variables, a chi-square or Fisher exact test was used, according to the frequency of observed events. Likewise, area under the curve (AUC) was calculated and a comparison by the chi-squared test was performed between the AUC of the NIH consensus risk stratification and the AUC of the classification based on volumetry. For all the statistical tests, a value of p ≤ .05 was established as the significance level. All statistical procedures were performed in STATA ver. 14.1 (StataCorp, Texas, USA) and were reviewed by a statistician.

Results

Basal Characteristics of the Patients

Of the 74 cases of GIST analyzed, 37 (50%) cases occurred in women. The median age was 58 (age range: 25 to 91 years). Half of the cases presented in the stomach. The tumor size had a median of 8 cm (range of 1–30 cm). Regarding the pathological characteristics, 60.5% cases were spindle cell, 30.2% were mixed, and 9.3% were epithelioid. A total of 11.6% cases showed skenoid fibers, 27.9% had intra-tumoral lymphoid aggregates, 4.7% cases had rhabdoid cells, 27.9% presented invasion to the mucosa with ulceration, and 23.3% presented marked pleomorphism. The median mitosis count for 50 HPF was 4 (range 0–92).

The patients had a median follow-up of 31 months (range 12–131 months) and during this period, disease recurred in 16 (21.6%). At the end of the study period, the 16 patients died.

Comparison of Groups with and Without Recurrence

We compared the clinical and pathological characteristics of patients with recurrent GIST (n = 16) with patients with non-recurrent GIST (n = 58), results of which are summarized in Table 2. The significant differences were that death of patients with recurrence was more and the follow-up period was longer. None of the other characteristics assessed was significantly different between the groups.

Table 2 Clinico-pathological characteristics of 74 patients with gastrointestinal stromal tumor (GIST) according to recurrence of the disease
Full size table

Comparison of Risk of Recurrence Classification Systems

The NIH risk of recurrence classification was compared with volumetry-based groups, with respect to the ability to discriminate the presence of recurrence. The results are described in Table 3. As Table 3 and Fig. 1 show, the AUC of the classification was superior to the NIH consensus, with statistical significance (p = .05). In a stratified analysis, the AUC of the volumetry in cases located at the stomach was 0.732, compared to 0.519 from the NIH classification (p = .037). For cases arising in other sites than the stomach, the AUC for the volumetry was 0.613 compared to 0.532 form the NIH classification (p = .047).

Table 3 Comparison of classifications for risk of recurrence in GIST
Full size table
Fig. 1
figure 1

Area under the curve (AUC) comparison between three classifications of the risk of recurrence grouping of GIST. a AUC of the National Cancer Institute consensus system of classification of GIST in four groups (AUC = 0.5743). b AUC of the classification in three categories using tumor volumetry in the surgical specimen in patients without adjuvant or neoadjuvant therapy (AUC = 0.6515). c AUC of the classification in three categories using tumor volumetry in the pre-surgical tomography in patients without adjuvant or neoadjuvant therapy (AUC = 0.6193)

Full size image

Discussion

In the present study, we show that volumetry could be a tool for stratifying in groups the recurrence risk of patients with GIST, independently of the site and without histopathology parameters. The stratification is based on volumetry, both in the surgical specimen and in tomography, and was superior to the NIH consensus classification in our series.

This has not been reported before, but some information about the usefulness of volumetry on GIST is available. Trumani et al. [3] previously reported that estimation of tumor volume in primary GIST using the mathematical formulae of ellipsoid volume is feasible, because GISTs are rarely spherical (and in cases where they are indeed spherical, we can use the sphere formula for volume calculation) and the segmented volumes were highly concordant with three axis-based scalene ellipsoid volumes. They found that this method is feasible, reproducible, and even comparable to the automated method (based on the Carestream Vue PACS Lesions Management Software, Carestream Health, Inc. N.Y.). In another report, Hashiba et al. [4] followed the growth of a small GIST and calculating their doubling time (which was 3.3 months); they deducted a high growth rate and malignancy. Thereafter, they performed a gastric resection and found that the GIST had 15–16 mitoses per 50 HPF, indicating malignancy. The patient was found to have hepatic metastasis 27 months after the surgery, confirming the malignant behavior of the tumor. Finally, volumetry has also been used to evaluate tumor response to medical treatment [5].

The most important risk factors for conventional GISTs are the anatomic site, size, and mitotic rate [6]. Other important risk factors are tumor rupture and mucosal invasion; however, true mucosal invasion is rare and subjective, so it is not now incorporated into the major risk stratification schemes for GIST [7]. Based on these parameters, several risk stratification schemes have been proposed. The first scheme that was established, the NIH Consensus Criteria: used mitotic rate and size to determine the risk of recurrence (Table 1) [2]. After it was established, its utility was confirmed in series with long-term follow-up [8, 9]. Based on several large studies, the Armed Forces Institute of Pathology (AFIP) modified the NIH Consensus criteria to add anatomic sites including the stomach, duodenum, jejunum/ileum, and rectum [10], and these criteria are recommended by the College of American Pathologists (CAP) [11]. Joensuu has proposed a simplification to the AFIP criteria that groups anatomic sites into either gastric or non-gastric sites, to show that gastric tumors have a better prognosis. Further, he subdivided mitotic rate into three categories instead of two: less than or equal to 5, 6–10 and greater than 10 mitotic figures per 50 HPF. Finally, he added tumor rupture as automatic criteria for determining a GIST as high risk.

Less complex classifications are needed because the major use of risk stratification criteria is for determining who should and should not receive adjuvant therapy after resection, and the ability to definitively find a few high-risk categories is helpful. Also, pre-surgical estimation of the risk of recurrence is valuable. We presented a simple and feasible classification in three categories, involving all tumor locations; however, we find some limitations. First, the follow-up time of the group of patients with no recurrence in our series was shorter than those with recurrence. Second, since the diameter evaluation was not performed in a systematic fashion by an expert radiologist, we have taken the diameter as settled in the radiology reports; however, this reflects the diary practice, where we have only the radiological report. Third, our series is small and our data needs to be validated in a larger series.

Conclusions

We found tumor volumetry by applying a mathematical formula, a simple, promising, and feasible method to classify into groups with risk of recurrence of primary GISTs, surgically treated and without administration of adjuvant therapy. This method was superior to the NIH consensus in predicting recurrence; however, due to the very selected and small sample size of our study, our results need to be validated.