Subsite‐specific trends in mid‐ and long‐term survival for head and neck cancer patients in Japan: A population‐based study

Abstract

Advances in diagnostic techniques and treatment modalities have impacted head and neck cancer (HNC) prognosis, but their effects on subsite‐specific prognosis remain unclear. This study aimed to assess subsite‐specific trends in mid‐ and long‐term survival for HNC patients diagnosed from 1993 to 2011 using data from population‐based cancer registries in Japan. We estimated the net survival (NS) for HNC by subsite using data from 13 prefectural population‐based cancer registries in Japan. Changes in survival over time were assessed by multivariate excess hazard model of mortality. In total, 68,312 HNC patients were included in this analysis. We observed an overall improvement in 5‐year NS for HNC patients in Japan. However, survival varied among subsites of HNC, with some, such as naso‐, oro‐ and hypopharyngeal cancers, showing significant improvement in both 5‐ and 10‐year NS, whereas others such as laryngeal cancer showed only a slight improvement in 5‐year NS and no significant change in 10‐year NS after adjustment for age, sex and stage. In conclusion, the study provides insights into changing HNC survival by site at the population level in Japan. Although advances in diagnostic techniques and treatment modalities have improved survival, these improvements are not shared equally among subsites.

We evaluated trends of head and neck cancer survival by subsite in Japan using population‐basedcancer registry data. During the observation periods, each pharyngeal cancer showed an upward trend, although laryngeal cancer showed no significant trend in long‐term survival. These findings may reflect the change in mainstream treatment.

Abbreviations

CI

confidence interval

HNC

head and neck cancer

NS

net survival

US

United States

1. INTRODUCTION

Head and neck cancers accounted for 29,000 new cases and 9000 deaths in Japan in 2019, ¹ comprising approximately 3% of all cancers arising in this country. The major subsites in Japan include the oral cavity, nasopharynx, oropharynx, hypopharynx, and larynx, which accounted for 32.6%, 3.9%, 13.5%, 14.0%, and 22.4% of cases, respectively. Although overall HNC incidence has increased over the past 20 years, trends vary by subsite: nasopharyngeal and laryngeal cancers have decreased, while oral cavity, oropharyngeal and hypopharyngeal cancers have increased. ² However, HNC survival trends by subsite are unknown yet.

Despite their development in anatomically close regions, each type of HNC has different risk factors, diagnostic and treatment approaches, and prognosis. ³ , ⁴ Smoking and drinking are major risk factors for all HNCs, ⁵ while viral infection contributes to cancers at specific subsites. In particular, human papillomavirus (HPV) and Epstein–Barr virus (EBV) have recently emerged as significant risk factors for oropharyngeal cancer and nasopharyngeal cancer, respectively. ⁶ , ⁷ The contribution of each risk factor varies by subsite, and as the prevalence of each individual risk factor changes, the trends in incidence differ for each subsite. Further, their impact on trends in survival is unknown.

Over the last few decades, two major advances have impacted the prognosis of HNC. First, diagnostic techniques have improved, including advances in imaging and endoscopic devices. ⁸ , ⁹ Second, treatment modalities for HNC patients have considerably progressed. For example, since the 1990s, clinical trials and meta‐analyses have demonstrated the efficacy and effectiveness of radiotherapy and chemoradiotherapy in the treatment of HNC while preserving laryngeal function, leading to these treatments becoming mainstream approaches, especially for pharyngeal and laryngeal cancer. ¹⁰ , ¹¹ , ¹² , ¹³ These advances in diagnostic technique and changes in treatment modality may have had an impact on the prognosis of HNC. Furthermore, they may have also had an impact on prognostic changes at individual subsites, ¹⁴ , ¹⁵ given that the effectiveness of advanced diagnostic and treatment methods varies by subsite.

Understanding trends in the survival of HNCs by subsite is crucial to the effective management of HNCs. While several studies in Western countries have shown a trend of improved survival, they also noted that this trend varied by primary tumor location. ¹⁴ , ¹⁵ , ¹⁶ For oral cavity and pharyngeal cancers, an upward trend in 5‐year survival has been observed in many countries. In contrast, several studies in the US indicated a downward trend in the 5‐year survival of laryngeal cancer. ¹⁷ , ¹⁸ , ¹⁹ For example, Hoffman et al. reported a decrease in 5‐year relative survival, from 68.1% for cases diagnosed in 1985 to 63.3% in 1996, ¹⁹ suggesting that the decline might be due to late toxicity from chemoradiotherapy. ²⁰ , ²¹ , ²² To our knowledge, however, no study has yet evaluated trends in survival by subsite in Asia, including Japan.

Here, we evaluated subsite‐specific trends in mid‐term (5‐year) and long‐term (10‐year) survival of HNC patients diagnosed between 1993 and 2011 using data from Japanese population‐based cancer registries.

2. MATERIALS AND METHODS

2.1. Data source

We used data from patients diagnosed with head and neck epithelial carcinoma from 13 population‐based cancer registries in Miyagi, Yamagata, Tochigi, Niigata, Fukui, Aichi, Shiga, Osaka, Tottori, Yamaguchi, Saga, Nagasaki, and Kumamoto. All of these registries were members of the Monitoring of Cancer Incidence in Japan (MCIJ) project in 2015. The 13 prefectures were selected from all 47 prefectural cancer registries in Japan participating in the MCIJ project, as follows (Figure S1). First, we selected 40 prefectural cancer registries whose data quality for all cancers in 2015 met the following standards of the MCIJ project 2015: (1) proportion of cases reported by death certificate only (DCO%: death certificate only) of less than 10%; (2) proportion of cases first notified through death certificate (DCN%: death certificate notification) of less than 20%; and (3) mortality to incidence ratio (M/I) of less than or equal to 0.5. From these, we further selected the 13 registries above because they submitted data between 1993 and 2011 to the MCIJ project in 2015. These selected registries encompassed data for 25.4% of the total population of Japan in 2015. ²³ They included information on patient characteristics, tumors, and treatment. Tumor localization and histology were recorded according to the International Classification of Disease for Oncology, Third Edition (ICD‐O‐3), with C00 codes for lip, C02‐06 for oral cavity, C07‐08 for salivary gland, C11 for nasopharynx, C01, 05, 09–10 for oropharynx, C12‐13 for hypopharynx, C14 for NOS, C30.0, C31 for nasal or paranasal, C30.1 for middle ear, and C32 for larynx. ²⁴ Epithelial malignancies (8000–8574) arising in the head and neck were included. Patients were followed up after diagnosis in all 13 prefectures, and linked to the prefectural death certificate database. Moreover, among them, residence records were used to confirm their vital status in eight prefectures (Miyagi, Yamagata, Tochigi, Fukui, Aichi, Shiga, Osaka, Nagasaki). Furthermore, only Osaka patients were tracked by residence records even if they moved to other prefectures. We set the final date of follow‐up as 31 December 2016.

2.2. Subject selection

We selected patients diagnosed with primary HNC from 1 January 1993 to 31 December 2011 from 13 prefectural cancer registries. In survival analysis, patients who met the following conditions were excluded: (1) registered by death certification only; (2) aged 14 years or younger or 100 years or older; (3) multiple cancers, of which HNC was not diagnosed first.

2.3. Outcome

The primary outcome of this study was survival after a cancer diagnosis. Observation was terminated on the date of death for any death that occurred within 5 or 10 years from diagnosis or was censored at 5 or 10 years from diagnosis if the patient survived.

2.4. Statistical analysis

We defined six periods as follows: period 1, 1993–1995; period 2, 1996–1998; period 3, 1999–2001; period 4, 2002–2004; period 5, 2005–2007 (2005–2006 for 10‐year survival); and period 6, 2008–2011. Age at diagnosis was classified into three groups (<65 years, 65–75 years, and > 75 years). Cancer stage was used in analyses by converting from the four registered categories to three categories: (1) localized (cancer limited to the original organ); (2) regional (cancer extension beyond the limit of the organ of origin); and (3) distant (metastasis to distant organs). Patient characteristics were described by the six periods. Trends in prognosis in HNC patients were assessed by NS analysis, which is useful in showing survival probability in the absence of other non‐cancer causes of death. Background mortality of HNC patients was calculated using the complete national population life tables by birth year, age, and sex. ²⁵ , ²⁶ NS was estimated using the net survival method of Pohar‐Perme. ²⁷ The 5‐ and 10‐year NS were calculated for patients diagnosed in each period for total HNC and each subsite. NS in each period was compared using the excess hazard model, a multivariate regression approach based on generalized linear models adopting the Poisson assumption for the observed number of deaths. ²⁸ The excess hazard ratio ²⁹ (EHR) was calculated with adjustment for the following covariates: age at diagnosis, sex, and cancer stage. We used the strs command in STATA to estimate NS, and the glm command to estimate EHR. ²⁶ We defined an improvement in NS as cases in which the p‐value for trend in EHR was less than 0.05.

Two‐sided p‐values <0.05 were considered to indicate statistical significance. All statistical analyses were conducted using STATA version 16.1 (StataCorp, College Station, TX, USA).

3. RESULTS

We identified a total of 68,312 patients with HNC as eligible for NS analysis, including 23,250 with oral cavity cancer, 2748 with nasopharyngeal cancer, 6979 with oropharyngeal cancer, 8413 with hypopharyngeal cancer and 17,214 with laryngeal cancer. We also identified a total of 44,454 deaths from HNC, including 14,067 due to oral cavity cancer, 1822 to nasopharyngeal cancer, 4895 to oropharyngeal cancer, 6733 to hypopharyngeal cancer, and 10,124 to laryngeal cancer in the 13 prefectural registries from 1993 through 2011. Patients are characterized by subsite and period in Table 1 and Table S1. Women accounted for 24.2% of total HNC, 39.7% of oral cavity cancer, 25.2% of nasopharyngeal cancer, 17.4% of oropharyngeal cancer, 10.3% of hypopharyngeal cancer, and 6.3% of laryngeal cancer. Proportions of older patients, aged 75 years or older at the time of diagnosis, steadily increased from period 1 to period 6 in major HNCs (oral cavity, pharynx, and larynx). According to cancer stage, the proportion of regional cancer patients in total HNC increased, and that of unknown decreased.

TABLE 1.

Characteristics of patients with oral, each pharyngeal and laryngeal cancer.

Primary site	Total		Period 1		Period 2		Period 3		Period 4		Period 5		Period 6
			1993–1995		1996–1998		1999–2001		2002–2004		2005–2007		2008–2011
	N	%	N	%	N	%	N	%	N	%	N	%	N	%
Total HNC	68,312	100.00	7454	100.00	8029	100.00	9103	100.00	10,014	100.00	12,590	100.00	21,122	100.00
Sex
Male	51,765	75.78	5773	77.45	6090	75.85	6904	75.84	7621	76.10	9560	75.93	15,817	74.88
Female	16,547	24.22	1681	22.55	1939	24.15	2199	24.16	2393	23.90	3030	24.07	5305	25.12
Age
15–64	29,670	43.43	3796	50.93	3897	48.54	4185	45.97	4438	44.32	5244	41.65	8110	38.40
65–74	21,636	31.67	2290	30.72	2535	31.57	2988	32.82	3164	31.60	4015	31.89	6644	31.46
75–99	17,006	24.89	1368	18.35	1597	19.89	1930	21.20	2412	24.09	3331	26.46	6368	30.15
Stage
Local	28,675	41.98	3197	42.89	3339	41.59	3685	40.48	3958	39.52	5220	41.46	9276	43.92
Regional	26,383	38.62	2482	33.30	2601	32.40	3167	34.79	3781	37.76	5126	40.71	9226	43.68
Distant	2705	3.96	324	4.35	304	3.79	318	3.49	430	4.29	494	3.92	835	3.95
Unknown	10,549	15.44	1451	19.47	1785	22.23	1933	21.23	1845	18.42	1750	13.90	1785	8.45
Oral cavity	23,250	100.00	2350	100.00	2733	100.00	3094	100.00	3396	100.00	4220	100.00	7457	100.00
Sex
Male	14,010	60.26	1516	64.51	1714	62.71	1850	59.79	2094	61.66	2484	58.86	4352	58.36
Female	9240	39.74	834	35.49	1019	37.29	1244	40.21	1302	38.34	1736	41.14	3105	41.64
Age
15–64	10,128	43.56	1277	54.34	1390	50.86	1450	46.86	1514	44.58	1674	39.67	2823	37.86
65–74	6548	28.16	614	26.13	744	27.22	911	29.44	934	27.50	1274	30.19	2071	27.77
75–99	6574	28.28	459	19.53	599	21.92	733	23.69	948	27.92	1272	30.14	2563	34.37
Stage
Local	11,068	47.60	1055	44.89	1203	44.02	1381	44.63	1494	43.99	2047	48.51	3888	52.14
Regional	8098	34.83	764	32.51	825	30.19	974	31.48	1200	35.34	1514	35.88	2821	37.83
Distant	540	2.32	80	3.40	78	2.85	78	2.52	90	2.65	83	1.97	131	1.76
Unknown	3544	15.24	451	19.19	627	22.94	661	21.36	612	18.02	576	13.65	617	8.27
Nasopharynx	2748	100.00	401	100.00	372	100.00	380	100.00	429	100.00	450	100.00	716	100.00
Sex
Male	2055	74.78	303	75.56	271	72.85	282	74.21	326	75.99	343	76.22	530	74.02
Female	693	25.22	98	24.44	101	27.15	98	25.79	103	24.01	107	23.78	186	25.98
Age
15–64	1692	61.57	274	68.33	233	62.63	226	59.47	267	62.24	275	61.11	417	58.24
65–74	684	24.89	92	22.94	96	25.81	103	27.11	106	24.71	113	25.11	174	24.30
75–99	372	13.54	35	8.73	43	11.56	51	13.42	56	13.05	62	13.78	125	17.46
Stage
Local	392	14.26	80	19.95	62	16.67	52	13.68	66	15.38	54	12.00	78	10.89
Regional	1689	61.46	200	49.88	221	59.41	231	60.79	258	60.14	301	66.89	478	66.76
Distant	227	8.26	32	7.98	21	5.65	19	5.00	43	10.02	33	7.33	79	11.03
Unknown	440	16.01	89	22.19	68	18.28	78	20.53	62	14.45	62	13.78	81	11.31
Oropharynx	6979	100.00	645	100.00	638	100.00	837	100.00	998	100.00	1352	100.00	2509	100.00
Sex
Male	5767	82.63	536	83.10	544	85.27	724	86.50	811	81.26	1132	83.73	2020	80.51
Female	1212	17.37	109	16.90	94	14.73	113	13.50	187	18.74	220	16.27	489	19.49
Age
15–64	3439	49.28	365	56.59	318	49.84	431	51.49	507	50.80	688	50.89	1130	45.04
65–74	2252	32.27	194	30.08	227	35.58	280	33.45	319	31.96	425	31.43	807	32.16
75–99	1288	18.46	86	13.33	93	14.58	126	15.05	172	17.23	239	17.68	572	22.80
Stage
Local	1505	21.56	164	25.43	153	23.98	205	24.49	212	21.24	245	18.12	526	20.96
Regional	4140	59.32	327	50.70	306	47.96	432	51.61	557	55.81	860	63.61	1658	66.08
Distant	382	5.47	35	5.43	42	6.58	46	5.50	59	5.91	67	4.96	133	5.30
Unknown	952	13.64	119	18.45	137	21.47	154	18.40	170	17.03	180	13.31	192	7.65
Hypopharynx	8413	100.00	688	100.00	790	100.00	1059	100.00	1257	100.00	1664	100.00	2955	100.00
Sex
Male	7545	89.68	600	87.21	690	87.34	944	89.14	1112	88.46	1501	90.20	2698	91.30
Female	868	10.32	88	12.79	100	12.66	115	10.86	145	11.54	163	9.80	257	8.70
Age
15–64	3677	43.71	373	54.22	398	50.38	503	47.50	566	45.03	715	42.97	1122	37.97
65–74	2967	35.27	204	29.65	255	32.28	370	34.94	447	35.56	603	36.24	1088	36.82
75–99	1769	21.03	111	16.13	137	17.34	186	17.56	244	19.41	346	20.79	745	25.21
Stage
Local	1645	19.55	108	15.70	147	18.61	198	18.70	207	16.47	339	20.37	646	21.86
Regional	5101	60.63	420	61.05	399	50.51	596	56.28	729	58.00	1025	61.60	1932	65.38
Distant	556	6.61	47	6.83	51	6.46	64	6.04	101	8.04	98	5.89	195	6.60
Unknown	1111	13.21	113	16.42	193	24.43	201	18.98	220	17.50	202	12.14	182	6.16
Larynx	17,124	100.00	2119	100.00	2232	100.00	2323	100.00	2494	100.00	3120	100.00	4836	100.00
Sex
Male	16,038	93.66	1989	93.87	2089	93.59	2186	94.10	2331	93.46	2943	94.33	4500	93.05
Female	1086	6.34	130	6.13	143	6.41	137	5.90	163	6.54	177	5.67	336	6.95
Age
15–64	6341	37.03	905	42.71	925	41.44	922	39.69	965	38.69	1110	35.58	1514	31.31
65–74	6415	37.46	813	38.37	858	38.44	890	38.31	919	36.85	1127	36.12	1808	37.39
75–99	4368	25.51	401	18.92	449	20.12	511	22.00	610	24.46	883	28.30	1514	31.31
Stage
Local	11,050	64.53	1349	63.66	1379	61.78	1392	59.92	1521	60.99	2008	64.36	3401	70.33
Regional	3423	19.99	381	17.98	412	18.46	434	18.68	507	20.33	669	21.44	1020	21.09
Distant	342	2.00	41	1.93	33	1.48	46	1.98	46	1.84	78	2.50	98	2.03
Unknown	2309	13.48	348	16.42	408	18.28	451	19.41	420	16.84	365	11.70	317	6.56

Abbreviation: HNC, head and neck cancer.

In total HNC, 5‐year NS improved by 6.2% and showed a significant trend in EHR, but 10‐year NS showed no significant trend (Figure 1A; Table 2). Survival varied among major HNCs (Table 2; Figure 1B–F). In oral cavity cancer, 5‐year and 10‐year NS showed an improving trend (Figure 1B; Table 2). The 5‐year NS was 60.0% (95% CI 57.6%–62.3%) and 67.1% (65.7%–68.4%) in periods 1 and 6, respectively (Table 2), while the 10‐year NS was 52.9% (49.8%–55.9%) and 55.2% (51.3%–58.9%) in periods 1 and 5, respectively. After adjustment for gender, age group, and cancer stage, 5‐year NS improved (p for trend <0.001), whereas 10‐year NS did not (Table 2).

FIGURE 1.

The 5‐year and 10‐year net survival rate of total HNC (A), oral cavity (B), nasopharyngeal (C), oropharyngeal (D), hypopharyngeal (E), and laryngeal cancer (F).

TABLE 2.

The 5‐year and 10‐year net survival and excess hazard ratio of patients with total HNC, oral cavity, and each pharyngeal and laryngeal cancer.

Primary site	Period	5‐year					10‐year
		NS	(95% CI)	EHR a	(95% CI)	p	NS	(95% CI)	EHR a	(95% CI)	p
Total HNC	1993–1995	57.2	(55.8–58.5)	1.00			50.7	(48.9–52.5)	1.00
	1996–1998	57.8	(56.5–59.1)	0.97	(0.91–1.02)	0.228	50.5	(48.8–52.3)	0.97	(0.92–1.02)	0.296
	1999–2001	58.7	(57.4–59.9)	0.95	(0.90–1.00)	0.056	50.3	(48.8–51.8)	0.96	(0.92–1.02)	0.170
	2002–2004	56.6	(55.5–57.8)	1.01	(0.95–1.06)	0.812	48.6	(47.0–50.2)	1.02	(0.97–1.07)	0.406
	2005–2007 b	60.2	(59.2–61.2)	0.89	(0.84–0.93)	4.32 × 10−6	49.2	(47.3–51.2)	0.94	(0.89–0.99)	0.025
	2008–2011	63.4	(62.6–64.2)	0.81	(0.77–0.85)	1.46 × 10−8
	p for trend					1.14 × 10−25					0.336
Oral cavity	1993–1995	60.0	(57.6–62.3)	1.00			52.9	(49.8–55.9)	1.00
	1996–1998	61.3	(59.0–63.5)	0.93	(0.84–1.03)	0.169	54.4	(51.6–57.2)	0.93	(0.85–1.03)	0.153
	1999–2001	59.8	(57.7–61.9)	0.98	(0.89–1.08)	0.717	54.2	(51.5–56.9)	0.97	(0.88–1.06)	0.520
	2002–2004	58.2	(56.1–60.1)	1.03	(0.94–1.14)	0.502	50.6	(47.7–53.4)	1.05	(0.96–1.14)	0.326
	2005–2007 b	64.7	(62.9–66.4)	0.81	(0.74–0.89)	1.69 × 10−5	55.2	(51.3–58.9)	0.87	(0.78–0.96)	0.005
	2008–2011	67.1	(65.7–68.4)	0.75	(0.69–0.82)	1.27 × 10−10
	p for trend					1.52 × 10−16					0.246
Nasopharynx	1993–1995	46.1	(40.8–51.4)	1.00			37.7	(31.8–43.5)	1.00
	1996–1998	46.4	(40.7–51.8)	0.94	(0.76–1.17)	0.594	35.3	(29.6–41.0)	0.97	(0.79–1.18)	0.741
	1999–2001	51.8	(46.2–57.1)	0.84	(0.67–1.04)	0.117	40.0	(34.2–45.7)	0.87	(0.71–1.06)	0.161
	2002–2004	51.6	(46.2–56.7)	0.88	(0.71–1.08)	0.231	40.3	(34.7–45.8)	0.88	(0.72–1.07)	0.200
	2005–2007 b	57.8	(52.5–62.6)	0.70	(0.56–0.87)	1.20 × 10−3	47.9	(40.7–54.9)	0.66	(0.52–0.83)	5.21 × 10−4
	2008–2011	60.1	(56.0–64.0)	0.64	(0.53–0.78)	1.07 × 10−5
	p for trend					3.30 × 10−7					7.90 × 10−4
Oropharynx	1993–1995	43.8	(39.4–48.1)	1.00			35.4	(30.4–40.5)	1.00
	1996–1998	41.7	(37.3–45.9)	1.07	(0.91–1.25)	0.411	31.7	(27.2–36.4)	1.06	(0.92–1.24)	0.415
	1999–2001	47.0	(43.2–50.8)	0.91	(0.78–1.07)	0.251	35.6	(31.6–39.6)	0.92	(0.80–1.06)	0.270
	2002–2004	46.0	(42.5–49.4)	0.93	(0.80–1.08)	0.334	36.9	(32.8–41.0)	0.89	(0.78–1.03)	0.117
	2005–2007 b	47.2	(44.2–50.2)	0.87	(0.76–1.00)	5.30 × 10−2	34.2	(30.1–38.3)	0.90	(0.78–1.04)	0.165
	2008–2011	55.0	(52.8–57.2)	0.73	(0.64–0.83)	1.83 × 10−6
	p for trend					6.10 × 10−11					0.021
Hypopharynx	1993–1995	24.6	(21.0–28.3)	1.00			19.5	(15.8–23.5)	1.00
	1996–1998	30.9	(27.4–34.5)	0.86	(0.76–0.98)	0.025	21.1	(17.6–24.9)	0.88	(0.78–1.00)	0.053
	1999–2001	34.5	(31.3–37.7)	0.78	(0.69–0.88)	8.28 × 10−5	25.9	(22.6–29.3)	0.80	(0.71–0.90)	2.04 × 10−4
	2002–2004	35.1	(32.1–38.1)	0.78	(0.69–0.88)	4.64 × 10−5	24.0	(20.7–27.5)	0.81	(0.72–0.91)	3.40 × 10−4
	2005–2007 b	41.0	(38.4–43.7)	0.64	(0.57–0.72)	4.03 × 10−14	27.3	(22.9–31.9)	0.69	(0.61–0.78)	3.59 × 10−9
	2008–2011	44.2	(42.1–46.2)	0.61	(0.54–0.68)	6.16 × 10−20
	p for trend					4.38 × 10−27					5.21 × 10−9
Larynx	1993–1995	74.5	(71.9–76.9)	1.00			69.4	(65.3–73.1)	1.00
	1996–1998	73.9	(71.3–76.3)	1.06	(0.91–1.24)	0.433	67.8	(63.6–71.7)	1.06	(0.92–1.22)	0.450
	1999–2001	76.2	(73.8–78.5)	0.96	(0.82–1.12)	0.597	66.1	(62.6–69.4)	1.01	(0.88–1.17)	0.868
	2002–2004	73.3	(70.9–75.5)	1.09	(0.94–1.26)	0.266	68.3	(64.8–71.5)	1.10	(0.96–1.26)	0.177
	2005–2007 b	75.5	(73.4–77.5)	0.98	(0.84–1.13)	0.762	63.4	(58.9–67.5)	1.07	(0.93–1.24)	0.347
	2008–2011	78.8	(77.1–80.4)	0.85	(0.74–0.97)	0.019
	p for trend					0.003					0.253

Abbreviations: EHR, excess hazard ratio of mortality; HNC, head and neck cancer; NS: net survival.

^a Adjusted by age at diagnosis, sex and cancer stage.

^b 2005–2006 in 10‐year analysis.

In naso‐, oro‐, and hypopharyngeal cancer, 5‐year and 10‐year NS showed an improving trend (Figure 1C–E). In the EHR model, both 5‐year and 10‐year NS showed improvements in NS, even after adjustment (Table 2).

In laryngeal cancer, 5‐year NS showed a slightly improving trend in the graph and EHR model. 5‐year NS was 74.5% (95% CI 71.9%–76.9%) and 78.8% (77.1%–80.4%) in periods 1 and 6, respectively (Figure 1F; Table 2). However, 10‐year NS showed a slightly worsening trend in the graph, at 69.4% (65.3%–73.1%) and 63.4% (58.9%–67.5%) in periods 1 and 5, respectively (Figure 1F). However, in the EHR model, the change was not significant (Table 2).

The 5‐year and 10‐year NS for lip, salivary gland cancer, nasal or paranasal cancer; middle year showed no statistically significant trend, while the middle ear had insufficient patients to calculate a valid NS (Tables S1 and S2; Figure S2).

4. DISCUSSION

In this study, we investigated trends in subsite‐specific HNC survival using large‐scale population‐based cancer registry data between 1993 and 2011 in Japan. We observed improved 5‐year and 10‐year NS for cancers of the nasopharynx, oropharynx, and hypopharynx over a period of approximately 20 years. For total HNC, oral cavity, and laryngeal cancers, 5‐year NS improved, whereas 10‐year NS did not. No improvement in NS was observed in cancers of the other subsites, namely lip and mid ear. To our knowledge, this is the first study to evaluate trends in HNC survival by subsite using data from population‐based cancer registries in Japan.

Overall, the 5‐ and 10‐year NS for HNCs showed improvement, suggesting that advances in medical practice played a role in this positive trend. Advances in diagnostic techniques, such as positron emission tomography detecting metastatic cancers and narrow‐band imaging in endoscopy detecting early‐stage or recurrent cancers, may have contributed to improved NS. ⁸ , ⁹ The adoption of chemoradiation as standard treatment ¹⁰ , ¹¹ , ¹² , ¹³ and the introduction of intensity‐modulated radiation therapy (IMRT) ³⁰ might have contributed to the improvement of NS, especially for patients with unresectable cancer. Due to the varying contributions of diagnostic techniques and changes in treatment modality for each cancer, there were differences in survival trends by subsite. For example, while the Introduction of IMRT contributed to improved survival in patients with nasopharyngeal cancer, ³¹ it did not have as significant an impact on improved survival in patients with cancers at other subsites. ³² , ³³

To distinguish whether advances in diagnostic techniques or treatment contributed to improved survival, we estimated long‐term survival rates separated by stage at diagnosis (Table S3). Even separated by cancer stage, 5‐year NS of total HNC and of oral, oropharyngeal, and hypopharyngeal cancers have significantly improved by periods for local and regional stages. With this additional result, we consider that the improved survival was due to advances in treatment rather than diagnostic techniques. If the improvement in survival occurred primarily by advances in diagnostic techniques, the percentage of local cancers detected early should increase. For total HNC, however, the proportion of local cancer did not increase significantly. Further, if advances in diagnostic technology cause stage migration only, overall survival for HNC should not have changed. However, the results of the additional analysis described above show that only local and regional survival had improved. This means that the improvement in overall survival was mainly due to advances in treatment for local and regional cancer.

In addition to the contribution to improved NS described above, other subsite‐specific factors significantly impacted the improvement of survival in major subsites of HNC. During the observation period, wider resection with the availability of free‐flap reconstruction and increasing administration of adjuvant radiotherapy and chemoradiotherapy might have contributed to the improvement of NS, especially the 5‐year NS, in oral cavity cancer. ³⁴ , ³⁵ The increasing proportion of HPV‐related oropharyngeal cancer cases might have contributed to improved survival, as these cases generally have a better prognosis than non‐infected cases. ³⁶ , ³⁷ , ³⁸ In hypopharyngeal cancer, the shift in treatment to a combination of surgery and adjuvant radiotherapy and chemotherapy might have contributed to the improvement in survival. ³⁹ , ⁴⁰

Oral cavity and laryngeal cancer did not show any improvement in 10‐year NS during the observation period. Surgical resection remains the primary treatment for oral cavity cancer ³⁵ and, while there have been advances in chemoradiotherapy, these might not have contributed as much to improving NS as they have for cancers in other sites. The shift from surgery to chemoradiation as a standard treatment for laryngeal cancer may have had a negative impact on the improvement in long‐term (10‐year) NS. This could be attributed to the increased risk of aspiration, ²¹ late toxicity, ²⁰ and difficulties in salvage surgery resulting in pharyngocutaneous fistula. ²² While chemoradiotherapy may be preferred by both patients and surgeons due to its positive impact on voice preservation and quality of life, it may not be the preferred choice for long‐term survival for laryngeal cancer.

Several previous studies have investigated the change in the 5‐year NS rate of HNC. ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ⁴¹ , ⁴² , ⁴³ However, results were inconsistent across different countries and study periods, even for the same subsite. For example, survival for laryngeal cancer showed a worsening in an early study in the USA, but no change in a recent study. Our findings on trends in survival for oral cavity, naso‐, oro‐, and hypopharyngeal cancer were consistent with those in several previous studies. ¹⁴ , ¹⁵ , ¹⁶ , ⁴¹ We saw no alignment with the pattern observed in two studies in the USA and Europe, ¹⁴ , ¹⁵ with our present finding that 5‐year NS for laryngeal cancer had improved whereas 10‐year NS had not improved.

This study has several strengths. First, it is the first to utilize high‐quality data from population‐based cancer registries in Japan to assess changes in the long‐term survival of HNC by subsite. By employing data from population‐based cancer registries, we were able to evaluate actual trends in head and neck cancer survival by subsite over a period of approximately 20 years until 2015, encompassing individuals with diverse backgrounds in the general population. Second, we used the excess hazard model to analyze trends in NS, taking account of changes in the distribution of sex, age, and stage. To our knowledge, no prior study has examined the impact of these factors on time trends in survival.

Several potential limitations of this study also warrant mention. First, the MCIJ dataset does not provide information on patient lifestyle, comorbidities, performance status, treatment history, or detailed pathological diagnosis such as HPV positivity/negativity. These factors may have an impact on patient outcomes, albeit the extent of any such influence remains unclear. Second, the Japanese population‐based cancer registries had issues with quality during the study period, especially early in the period, and failed to meet data quality for international standards. When cancer patients with poor prognoses are not registered or patients are not appropriately followed up, estimations will be biased and survival might be overestimated. In this study among data from all prefectures in Japan, we selected only 13 prefectures with high‐quality data for our analysis to minimize bias in outcome to the degree possible. In addition, data quality improved during the study period, especially with the development of the hospital‐based cancer registry beginning in the early 2000s, which may have led to the underestimation of stable or increasing trends in survival. Third, we were only able to analyze tracking data through 2016. We consider that the trends shown in this study will continue, as no breakthrough treatment changes have occurred since Cetuximab was introduced in 2012. In addition, subjects with a diagnosis of HNC followed by the diagnosis of a second cancer could not be differentiated from subjects with HNC only because the information used in this study was provided to us in an anonymized form. The occurrence of subsequent cancer could have significantly impacted the prognosis.

In conclusion, this study provides valuable insights into the changing survival trends of HNC patients at the population level in Japan. While advances in diagnostic techniques and treatment modalities may have led to overall improvements in survival, these improvements vary significantly by subsite. Understanding these subsite‐specific trends is crucial to better management of HNC and the creation of treatment strategies aimed at optimizing patient outcomes. Further research is needed to explore the reason behind the variations in survival trends and to develop specialized strategies aimed at addressing these disparities.

AUTHOR CONTRIBUTIONS

Hiroshi Tsuge: Data curation; formal analysis; investigation; visualization; writing – original draft; writing – review and editing. Daisuke Kawakita: Writing – original draft; writing – review and editing. Yukari Taniyama: Software; writing – review and editing. Isao Oze: Writing – review and editing. Yuriko N. Koyanagi: Writing – review and editing. Megumi Hori: Resources; writing – review and editing. Kayo Nakata: Resources; writing – review and editing. Hiromi Sugiyama: Resources; writing – review and editing. Isao Miyashiro: Resources; writing – review and editing. Izumi Oki: Resources; writing – review and editing. Yoshikazu Nishino: Resources; writing – review and editing. Kota Katanoda: Resources; writing – review and editing. Yuri Ito: Methodology; resources; software; writing – review and editing. Akiko Shibata: Resources; writing – review and editing. Tomohiro Matsuda: Resources; writing – review and editing. Shinichi Iwasaki: Writing – review and editing. Keitaro Matsuo: Conceptualization; writing – review and editing. Hidemi Ito: Conceptualization; funding acquisition; project administration; resources; software; supervision; writing – original draft; writing – review and editing.

CONFLICT OF INTEREST STATEMENT

Keitaro Matsuo is an editorial board member; other authors have no conflicts of interest.

ETHICS STATEMENT

Approval of the research protocol by an Institutional Reviewer Board: N/A.

Informed Consent: N/A.

Registry and the Registration No. of the study/trial: N/A.

Animal Studies: N/A.

Supporting information

Figure S1.

Figure S2.

Table S1.

Table S2.

Table S3.

Tsuge H, Kawakita D, Taniyama Y, et al. Subsite‐specific trends in mid‐ and long‐term survival for head and neck cancer patients in Japan: A population‐based study. Cancer Sci. 2024;115:623‐634. doi: 10.1111/cas.16028