Radiological Knosp, Revised-Knosp, and Hardy-Wilson Classifications for the Prediction of Surgical Outcomes in the Endoscopic Endonasal Surgery of Pituitary Adenomas: Study of 228 Cases

Abstract

Purpose: To evaluate which radiological classification, Knosp, revised-Knosp, or Hardy-Wilson classification, is better for the prediction of surgical outcomes in the endoscopic endonasal transsphenoidal (EET) surgery of pituitary adenomas (PAs).

Methods: This is a retrospective study of patients with PAs who underwent EET PA resection for the first time between January 2009 and December 2020. Radiological cavernous sinus invasiveness was defined as a Knosp or revised-Knosp grade >2 or a grade E in the Hardy-Wilson classification.

Results: A total of 228 patients with PAs were included. Cavernous sinus invasion was evident in 35.1% and suprasellar extension was evident in 74.6%. Overall, surgical cure was achieved in 64.3% of patients. Surgical cure was lower in invasive PAs than in non-invasive PAs (28.8% vs. 83.1%, p < 0.0001), and the risk of major complications was higher (13.8% vs. 3.4%, p = 0.003). The rate of surgical cure decreased as the grade of Knosp increased (p < 0.001), whereas the risk of complications increased (p < 0.001). Patients with Knosp 3B PAs tended to achieve surgical cure less commonly than Knosp 3A PAs (30.0% vs. 56.0%, p = 0.164). Similar results were observed based on the invasion and extension of Hardy-Wilson classification (stage A-C 83.1% vs. E 28.8% p < 0.0001, grade 0-II 81.1% vs. III-IV 59.7% p = 0.008). The Knosp classification offered the greatest diagnostic accuracy for the prediction of surgical cure (AUC 0.820), whereas the invasion Hardy-Wilson classification lacked utility for this purpose (AUC 0.654).

Conclusion: The Knosp classifications offer a good orientation for the estimation of surgical cure and the risk of complications in patients with PAs submitted to EET surgery. However, the invasion Hardy-Wilson scale lacks utility for this purpose.

Keywords: Hardy-Wilson classification; Knosp classification; endoscopic endonasal transsphenoidal surgery; invasive pituitary adenomas; pituitary adenomas.

Figure 1

Study population.

Figure 2

Hardy–Wilson and Knosp classifications. Hardy–Wilson classification considered the degree of sellar destruction: Grade 0 when the adenoma remains within the anatomical confines of the osteoaponeural sheath of the sella turcica; Grade I: the sella turcica is within normal limits in size or focally expanded and the tumor is <10 mm; Grade II: tumor ≥ 10 mm and the sella turcica is enlarged but the floor remains intact; Grade III: a local erosion or destruction of the floor; Grade IV when the entire floor of the sella is diffusely eroded or destroyed. Extrasellar extension according to Hardy–Wilson is divided in stage 0, with no suprasellar extension, A–C for progressive suprasellar extension. Knosp–Steiner classification considered: Knosp 0 when PA is medial to medial tangent; Knosp 1 if PA extends to the space between the medial tangent and the intercarotid line; Knosp 2 when PA extends to the space between the intercarotid line and the lateral tangent; Knosp 3 if PA extends lateral to the lateral tangent; and Knosp 4 with a complete encasement of intracavernous ICA. Knosp score 3–4 were considered as invasive PA. The revised-Knosp classification includes Knosp 3A when PA is above the intracavernous ICA into the superior cavernous sinus compartment and Knosp 3B when PA is below the intracavernous ICA into the inferior cavernous sinus compartment.

Figure 3

Intraoperative–radiologic correlation of cavernous sinus invasion in pituitary adenomas. ICA is highlighted in yellow dotted line and differently affected CS compartments are pointed out (white stars). Case 1: Right superior compartment invasion of the cavernous sinus (Knosp 3A) in an acromegalic patient. Preoperative MRI (A1) and intraoperative view through a 45° endoscope (A2) after tumor resection (left cavernous sinus medial wall resected in blue dotted line). The patient was cured after surgery. Case 2: Left Inferior compartment invasion of the CS in a resistant prolactin-secreting PA (Knosp 3B). Preoperative MRI (B1) and intraoperative view through a 0° endoscope (B2) after tumor resection showing anterior CS wall resection (green dotted line). Case 3: Complete cavernous sinus invasion (Knosp 4) in an acromegalic patient. Preoperative MRI in T2 sequences (C1, C2) shows a complete ICA encasement and the postoperative coronal MRI (C3) shows a near total resection. Intraoperative view through a 45° endoscope (C4) after tumor resection showing ICA and superior and inferior compartment. No surgical cure was achieved.

Figure 4

Surgical remission in functioning pituitary adenomas and total resection in non-functioning pituitary adenomas based on Knosp and Hardy classifications.

Figure 5

Diagnostic accuracy of the Knosp and Hardy scales for the prediction of surgical cure. (a1) AUC of the Knosp scale for the prediction of surgical failure: 0.820 [0.760–0.879]; optimal cutoff for the prediction of failure: Knosp 3 [sensitivity (Se) = 70.4% (59.7–79.2), specificity (Sp) = 84.2% (77.5–89.3); a2] AUC of the revised-Knosp scale for the prediction of surgical failure: 0.820 [0.760–0.882]; optimal cutoff for the prediction of failure: Knosp 3A [Se= 68.8% (57.8–78.1), Sp = 84.8% (78.1–89.8); a3] AUC of the invasion Hardy scale for the prediction of surgical failure: 0.654 [0.580–0.728]; optimal cutoff for the prediction of failure: Hardy IV [Se = 86.7% (75.8–93.1), Sp = 31.9% (24.0–40.9); a4] AUC of the extension Hardy scale (A–D vs. E) for the prediction of surgical failure: 0.732 [0.672–0.793]; optimal cutoff for the prediction of failure: Hardy stage E [Se = 62.2% (51.4–71.9), Sp = 84.2% (77.5–89.3)].