Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the gastrointestinal (GI) tract, and arise from interstitial cells of Cajal (ICCs) or their precursor cells (1–5). GISTs tend to form well-circumscribed, protruding nodules without diffuse infiltration, so that local resection is often initially an adequate therapy (3,6–8). A 1–2 cm surgical safety margin is thought to be required for grossly and microscopically complete resection of GISTs (8–10), referred to as an R0 resection in the Union for International Cancer Control (UICC) system (11). Novitsky et al (9) demonstrated that all 50 surgically removed GISTs with a grossly 1–2 cm margin beyond the tumors exhibited microscopic negative margins ranging from 0.2 to 4.5 cm. Based on this, Everett et al (10) deduced a possible 0.5-cm length microscopic extension of GISTs. To the best of our knowledge, however, detailed histopathological examination of such laterally extending or spreading lesions of GISTs has not been previously conducted. Therefore, we examined the incidence and the histopathological features of these laterally spreading lesions of GISTs in the present study.

Table I summarizes the clinicopathological findings of the 52 GISTs. Of the 52 GISTs, 29 (56%) were exophytic, 8 (15%) were pedunculated, 8 (15%) were dumb-bell shaped, 13 (25%) exhibited ulceration, 43 (83%)/9 (17%) were spindle-cell/epithelioid-cell predominant, 13 (25%) had a high mitotic rate (>5/50 high-power fields), 14 (27%) exhibited mucosal invasion, and 8 (15%) contained skeinoid fibers. Focal SMA and S-100 positivity was found in 14 (27%) and 4 (8%) GISTs, respectively. In 38 GISTs immunostained for DOG1, 37 (97%) exhibited positivity. Eleven (21%) and 10 GISTs (19%) were assessed as intermediate and high risk, respectively. Laterally spreading features were identified in 7 GISTs (13%). In 2 GISTs, the laterally spreading tumor cells involved the surgical margin, and were referred to as an R1 resection according to the UICC system (11). No other R1 cases were found.

A summary of the clinicopathological findings of laterally spreading GISTs is provided in Table II. Of these, 1 arose in the jejunum, while the other 6 were present on the gastric walls, including the posterior wall of fundus (2 cases), corpus (3 cases) and antral anterior wall (1 case). Five (71%) were exophytic, while 4 exhibited pedunculated features. Endoscopically, submucosal tumor-like protrusions or mildly elevated features were observed in 4 GISTs (Fig. 1A; cases 1, 3, 5 and 6) and concomitant ulceration was found in one (case 6). However, these endoscopic features did not appear to be different from those of other GISTs without laterally spreading lesions. The other 3 laterally spreading GISTs (cases 2, 4 and 7) exhibited no elevated tumorous features endoscopically (Fig. 1B and C), although one of them (case 7) had well-demarcated, depressed lesions (Fig. 1C). Histologically, the laterally spreading features were not found in the mucosa or submucosa, and were localized within the muscularis propria without significant thickening adjacent to the main GISTs (Figs. 2–4). The lengths of these spreading lesions ranged from 0.12 to 0.7 cm (mean, 0.3 cm). In the 2 R1 resection cases, the removed muscular layers adjacent to the main GISTs were relatively limited, and their surgical margins were involved by the spreading lesions (Fig. 2C and F). These spreading cells histologically resembled those of tumor cells of the main GISTs in 4 cases (Figs. 2G and H, and 3B, F and I). In another 3 cases, the spreading cells consisted of more slender spindle cells with smaller nuclei, compared with tumor cells of the main GISTs (Fig. 4D and F). Compared with immunohistochemical features of the main GISTs, KIT⁺ and DOG1 staining of the spreading lesions were similar in 4 cases (Fig. 4E and G), but were weaker or diminished in the other 3 cases (Fig. 2G, inset and H, inset). There were no differences of CD34⁺ staining between the main GISTs and the spreading lesions (Figs. 2E and 4C). No SMA or S-100 positivity was found in the spreading spindle cells. One patient (case 6) succumbed to the disease 2.5 years after the surgery. In this case, the autopsy revealed a 5.5-cm recurrent growth of the GIST at the gastric excision site and multinodular peritoneal seeding in the upper abdominal cavity. The other 6 patients with laterally spreading GISTs were alive 0.4–19.2 years after the surgery, with no evidence of disease.

The laterally spreading features were significantly associated with pedunculated GISTs (P=0.006), but not with older age (P=0.312), sex (P=0.969), tumor location (gastric or non-gastric) (P=0.725), tumor size (P=0.430), exophytic type (P=0.626), dumb-bell shape (P=0.634), ulceration (P=0.977), microscopic type (spindle- or epithelioid-cell predominant) (P=0.757), high mitotic rate (P=0.815), mucosal invasion (P=0.666), skeinoid fibers (P=0.634), or higher risk category (high and intermediate risk category) (P=0.872) (Table III).

The present study revealed unique laterally spreading features in 13% of a total of 52 GISTs. These spreading lesions were harbored within the otherwise normal-looking adjacent muscularis propria, so that recognition would be challenging either radiologically, endoscopically, or surgically. In fact, the present study did not reveal endoscopic features or signs specific to the laterally spreading lesions. These spreading lesions were not extensive (range, 0.12–0.7 cm), but infrequently (29% of the 7 cases) involved the muscular surgical margins. These findings not only support the generally accepted concept that R0 resection requires a 1 to 2-cm safety margin (8–10), but also indicate that such safety margins should be applied to the muscularis propria, not the mucosa or submucosa. In addition, the present study demonstrated a close relationship between laterally spreading lesions and the pedunculated shape of GISTs. Therefore, surgeons should pay attention to the ≥1-cm muscular safety margins for R0 resection, particularly for pedunculated GISTs, even if these muscular walls are not thickened or tumorous.

In some cases, laterally spreading tumor cells were somewhat different from those of the main tumors. Compared with the pathological features of the main GISTs, laterally spreading lesions consisted of more slender tumor cells with smaller nuclei (in 3 cases), and showed weaker or diminished KIT/DOG1⁺ staining (in 3 cases). However, the former 3 cases were not identical to the latter 3 cases, and the slender morphological features of laterally spreading tumor cells did not appear to be associated with the weaker or diminished KIT/DOG1⁺ staining. On the other hand, the more slender morphological features of laterally spreading tumor cells somewhat resembled those of ICC hyperplasia (4,6,13). It is known that ICCs can be positive for DOG1 (14,15). These findings raise the possibility that some laterally spreading lesions may represent a hyperplastic reaction of ICCs caused, secondarily, by torque from the stalk of the pedunculated GISTs. However, the other 4 spreading lesions identified were composed of tumor cells resembling those of the main GISTs, suggesting true intramuscular extension of the tumor cells.

The most important risk factors associated with GISTs are anatomic location, tumor size, and mitotic activity, all of which are included in the widely accepted risk-stratification criteria (1–6,12). Tumor rupture is another independent risk factor of GISTs, despite its rarity (3–7,12). Mucosal invasion, high cellularity, and increased microvessel density may be additional indicators of poor prognosis (1,3,4,6). Some authors noted high recurrence or poor prognosis in incomplete resection cases of GISTs compared with complete resection cases (8,16). However, DeMatteo et al (17) concluded that microscopically positive surgical margins did not influence patient outcomes. Zhi et al (18) demonstrated that a microscopically positive surgical margin can impact the disease-free survival of patients with GISTs, but had no influence on overall survival. Therefore, the prognostic significance of incomplete GIST resection remains controversial (6,9,10,19). In the present study, both patients that had laterally spreading-related R1 resection were alive without recurrence 1–2.3 years after surgery, demonstrating that they were not associated with unfavorable outcomes. However, the recurrence or metastases of GISTs may occur many years later (6), so that further follow-up is required. In addition, one patient with a laterally spreading GIST succumbed to the disease 2.5 years after surgery, although the surgical margin in this case was free of tumor cells and the spreading lesions exhibited ICC hyperplasia-like slender morphology. Therefore, regarding the clinicopathological significance of the laterally spreading lesions, further investigations of a larger series of GISTs are needed.

To the best of our knowledge, the present study is the first to histologically describe the laterally spreading features of GISTs, which occurred in 13% of the GISTs evaluated. These spreading lesions may contribute to R1 resection, albeit uncommonly. For R0 resection, the ≥1-cm muscular safety margin should be required, particularly in cases of pedunculated GISTs.