Localization patterns and survival of extranodal NK/T-cell lymphomas in the United States: A population-based study of 945 cases

Extranodal natural killer/T-cell lymphomas (eNK/TCLs) are aggressive, angioinvasive, necrotizing EBV-associated lymphomas with a 5-year overall survival ranging from 30–50%.¹ Approximately 80% of eNK/TCLs develop in the nose/upper aerodigestive tract and 20% at non-nasal sites (skin, viscera, lymph nodes [LN]).^1,2 Mean age of onset is 44 years, and it is 2–3 times more common in men.³

To further investigate the impact site of initial diagnosis on prognosis, we examined data from the Surveillance, Epidemiology, and End Results Program (SEER-18) database for patients diagnosed with eNK/TCL from 2000–2016 (n=945). We analyzed demographics and survival based on the site of initial diagnosis. Localized lymph node involvement was defined as within a single lymph node region, while regional involved two or more regions on the same side of the diaphragm, and distant, regions on both sides of the diaphragm. Paired t-test and chi-square tests were used with alpha <0.0025 with the Bonferroni correction. Kaplan-Meier analyses were performed in R. Cox multivariate regression models were performed with respect to age, gender, location at diagnosis, and stage.

The median age of diagnosis of eNK/TCL was 53.5 years (range 3–103). The median age of patients with skin as primary site of diagnosis (62 years [range 15–103]) was significantly older than those diagnosed at other sites: oral/nasal (54 years [range 4–93]), visceral (54.5 years [range 3–96]), or LN (51.5 years [range 3–87])(p<0.05 for all comparisons with skin, Table 1).

Table 1.

Population demographics and Cox regression model.

		All Sites n=936	Oro/Nasal n=678	Visceral n=102	Lymph Node n=104	Skin n=52	p-values
Distribution			678 (72.4%)	102 (10.9%)	104 (11.1%)	52 (5.6%)
Age
Median (range) years		54 (3–103)	54 (4–93)	54.5 (3–96)	51.5 (3–87)	62 (15–103)	Skin vs. all others p<0.0001
Gender
	Male	602 (64.3%)	432 (63.7%)	69 (67.6%)	72 (69.2%)	29 (55.8%)	Skin vs. all others p<0.0001
	Female	334 (35.7%)	246 (36.3%)	33 (32.4%)	32 (30.8%)	23 (44.2%)
Race
	Hispanic (all races)	335 (35.8%)	254 (37.5%)	33 (32.4%)	37 (35.6%)	11 (21.2%)	NS
	Non-Hispanic American Indian/Alaska Native	11 (1.2%)	6 (0.9%)	4 (3.9%)	1 (1.0%)	0 (0.0%)
	Non-Hispanic Asian or Pacific Islander	202 (21.6%)	151 (22.3%)	23 (22.5%)	16 (15.4%)	12 (23.1%)
	Non-Hispanic Black	46 (4.9%)	32 (4.7%)	6 (5.9%)	7 (6.7%)	1 (1.9%)
	Non-Hispanic White	339 (36.2%)	232 (34.2%)	36 (35.3%)	43 (41.3%)	28 (53.8%)
	Non-Hispanic Unknown Race	3 (0.3%)	3 (0.4%)	0 (0.0%)	0 (0.0%)	0 (0.0%)
Stage
	Localized	405 (45.9)	345 (53.0)	31 (33.7)	5 (5.3)	24 (53.3)	P<0.0001 for all
	Regional	190 (21.6)	162 (24.9)	13 (14.1)	13 (13.9)	2 (4.4)
	Distant	286 (32.4)	143 (22.0)	48 (52.2)	76 (42.2)	19 (42.2)
Survival
	Median follow-up (range), months	10 (0–190)	14 (0–187)	4 (0–190)	4 (0–175)	9.5 (0–172)	NS
	# Alive (%) at end of follow-up	375 (40.1%)	326 (48.1%)	26 (25.5%)	16 (15.4%)	7 (13.5%)	Skin vs. oro/nasal p<0.00001 Skin vs. Visceral p=0.085 Skin vs. LN p=0.75
	# Dead (%) at end of follow-up	561 (59.9%)	352 (51.9%)	76 (74.5%)	88 (84.6%)	45 (86.5%)
	2-year OS (95% CI) (Non-age adjusted)	0.46 (0.42, 0.49)	0.61 (0.55, 0.69)	0.26 (0.19, 0.37)	0.26 (0.19, 0.36)	0.24 (0.15, 0.39)	Oro/Nasal vs. all others p<0.0001 Skin vs. LN and visceral NS
	5-year OS (95% CI) (Non-age adjusted)	0.37 (0.34, 0.41)	0.53 (0.50, 0.57)	0.24 (0.16, 0.34)	0.18 (0.12, 0.28)	0.17 (0.09, 0.32)	Oro/Nasal vs. all others p<0.0001 Skin vs. LN and visceral NS
	2-year DSS (95% CI) (Non-age adjusted)	0.55 (0.51, 0.58)	0.62 (0.58, 0.66)	0.34 (0.25, 0.47)	0.35 (0.27, 0.47)	0.36 (0.23, 0.54)	Oro/Nasal vs. all others p<0.0001 Skin vs. LN and visceral NS
	5-year DSS (95% CI) (Non-age adjusted)	0.47 (0.44, 0.51)	0.54 (0.49, 0.58)	0.32 (0.23, 0.45)	0.26 (0.18, 0.38)	0.32 (0.20, 0.51)	Oro/Nasal vs. all others p<0.0001 Skin vs. LN and visceral NS
Cox Regression Model
	Disease-Specific Survival		Overall Survival
Covariate	(95% CI)	p-value	(95% CI)	p-value
Site
Skin vs. oro/nasal	1.55 (1.06, 2.27)	0.023	1.63 (1.19, 2.25)	0.003
LN vs. oro/nasal	1.57 (1.17, 2.09)	0.002	1.50 (1.16, 1.94)	0.002
Visceral vs. oro/nasal	1.65 (1.23, 2.21)	<0.001	1.65 (1.27, 2.13)	<0.001
LN vs. skin	1.01 (0.65, 1.55)	0.971	0.92 (0.63, 1.33)	0.654
Visceral vs. skin	1.06 (0.68, 1.65)	0.783	1.01 (0.69, 1.47)	0.970
Age (10 years)	1.09 (1.04, 1.15)	<0.001	1.15 (1.09, 1.20)	<0.001
Sex, Male vs. Female	0.85 (0.70, 1.04)	0.107	0.98 (0.82, 1.17)	0.819
Stage
Regional vs. localized	1.77 (1.33, 2.33)	<0.001	1.66 (1.31, 2.11)	<0.001
Distant vs. localized	3.25 (2.53, 4.17)	<0.001	2.74 (2.21, 3.41)	<0.001
Unknown/Unstaged vs localized	1.18 (0.72, 1.93)	0.513	1.60 (1.11, 2.31)	0.011

Patients with initial diagnosis at oral/nasal, visceral, or LN sites were more likely to be male than patients diagnosed with disease in the skin (63.6%, 67.6% and 69.2% male, respectively, versus 55.8% for skin disease).

Analysis revealed significantly poorer prognosis for patients with skin-diagnosed eNK/TCL compared with oro/nasal disease. Five-year overall survival rates were 61% (95% confidence interval [CI] 55–69%) for oral/nasal-diagnosed disease, while LN-diagnosed, viscerally-diagnosed, and skin-diagnosed had 5-year survival rates of 18% (95% CI 12–28%), 17% (95% CI 9–32%) for skin-diagnosed, and 24% (95% CI 16–34%), respectively (Figure 1). Patients with skin-diagnosed disease were more likely to have advanced stage disease than those with oro/nasal disease. Thus, it appears that skin-diagnosed disease behaves more like lymph node or viscerally involved disease than oral/nasal disease. This may be a reflection of skin, LN, and visceral involvement being late sequelae of the disease.

Figure 1.

Overall survival by location of extranodal NK/T-cell lymphoma (eNKTCL). Cases diagnosed based on oro/nasal involvement demonstrate better survival than those diagnosed at lymph node, skin, or visceral sites.

Cox multivariate regression modelling revealed that patients with skin-diagnosed disease were 1.56 times (95% CI 1.17–2.09, p=0.023) more likely to die of eNK/TCL than those with oral/nasal diagnoses. This was similar to lymph node diagnosed disease (relative risk [RR] 1.57, 95% CI 1.17–2.09, p=0.002) and visceral disease (RR 1.65, 95% CI 1.23–2.21, p<0.001). Age was a significant factor, with a 9% increase risk of death due to eNK/TCL per decade of increased age (RR 1.09, 95% CI 1.04–1.15, p<0.001). Extent of disease was also significant, localized and regional disease being protective factors compared to distant disease. Gender was not a significant contributor.

In this study, patients with eNK/TCL diagnosed in the skin had marked lypoor prognosis comparable to that of patients diagnosed initially in the oral/nasal area, likely due to skin involvement being associated with advanced disease. Survival of patients with skin-diagnosed disease was similar to patients diagnosed via visceral or lymph node disease. These findings offer valuable prognostic information for patients and can help stratify survival upon diagnosis, even prior to other systemic work-up.