Table 2.

Overview of studies on NTCP models for moderate-to-severe xerostomia as assessed by the EORTC QLQ-H&N35.^a

Reference	Time point, mo	Quality criteria for NTCP models										Level of evidence
		RT technique	Study design	Analysis	No. of patients	No. of events	AUC	Calibration tested	Internal validation	External validation	Individual NTCP estimation possible
Current study	6	Prim IMRT	Prospective	External validation	669	338	0.72	Yes	Yes	Yes	Yes	2a
Beetz 2012a [10]	6	Prim 3DCRT	Prospective	Multivariable	165	86	0.82	No	Yes	No	Yes	3
Beetz 2012b [9]	6	Prim IMRT	Prospective	External validation	161	83	0.66	Yes, poorb	Yes	Poor	Yes	3
Beetz 2012c [8]	6	Prim IMRT	Prospective	Multivariable	161	83	0.69	Yes	Yes	No	Yes	3
Lee 2012 [11]	12	Prim + PO IMRT	Prospective	Univariable	31	NM	0.75	Yes	No	No	Yes	4
Lee 2013 [12]	3	Prim + PO IMRT	Prospective	LKB-model	237	89	0.68	No	No	No	Yes	4
	12				146	49	0.73	No	No	No	Yes	4
Lee 2014 [13]	3	Prim + PO IMRT	Prospective	Multivariable	185	87	0.86	No	Yes	No	Yes	3
	12				117	43	0.86	No	Yes	No	Yes	3
												3
Jellema 2005 [14]	6	Prim + PO 3DCRT	Prospective	Multivariable	113	46	No	No	No	No	No	4
	12				94	35	No	No	No	No	No	4

Abbreviations: NTCP, Normal Tissue Complication Probability; EORTC QLQ-H&N35, European Organization for Research and Treatment Quality of Life Questionnaire Head and Neck 35; RT, radiation therapy; AUC, Area Under the Curve.; Prim, primary; IMRT, intensity-modulated radiation therapy; 3DCRT, 3-dimensional conformal radiation therapy; PO, postoperative; LKB, Lyman-Kutcher-Burman model.

^aThe NTCP model of Beetz 2012c [8] was ranked highest and subjected to external validation as described in this article.

^bThis study tried to validate the model of Beetz 2012a [10], developed in patients treated with 3DCRT, in a patient population treated with IMRT. This model performed poorly in the IMRT population, especially regarding calibration.