Histopathological grading of adenoid cystic carcinoma of the head and neck: analysis of currently used grading systems and proposal for a simplified grading scheme

Background: Histopathological grading of adenoid cystic carcinoma (ACC) is a controversial issue. It is generally agreed that solid type ACC has a relatively poor prognosis. However, the amount of solid regions within this often mixed type tumor that predicts a poor prognosis is not firmly established. Some authors stipulate that the presence of a solid component regardless of the amount is a poor prognosticator where others argue that the amount should be taken into consideration. Two grading systems most commonly used are those described by Perzin et al./Szanto et al. and Spiro et al., respectively. They report that prognosis of ACC is poor if >30% and >50% of the tumor volume has a solid growth pattern, respectively.

Material and methods: The described grading systems are applied to a series of 81 surgically treated cases of ACC at the VU University Medical Center, Amsterdam, The Netherlands. Moreover, we introduced an alternative grading system, in which the presence of a solid component, irrespective of its amount, is considered. All three systems of grading were tested for inter-observer concordance and prediction of prognosis.

Results: Inter-observer concordance for grading ACC according to Perzin et al./Szanto et al. and Spiro et al., proved to be moderate with Kappa Scores of 0.393 and 0.433, respectively. Our alternative grading system yielded inter-observer concordance with a Cohen's kappa result of 0.990. All systems were comparable in discriminating patients with poor clinical outcome. Histopathological grade proved to be an independent prognosticator.

Conclusion: The presence of any solid component in ACC is a negative prognosticator, and can histopathologically be diagnosed with a high reliability. These results suggest to merely register the presence or absence of a solid tumor component since its inter-observer variability is very low, its reproducibility is high and its predictive value is comparable to the traditional grading systems used.