Clinical Outcomes and Prognostic Factors of Adenoid Cystic Carcinoma of the Head and Neck

Abstract

Adenoid Cystic Carcinoma (ACC) is a salivary gland malignancy with unpredictable growth and poorly understood prognostic factors. A database search of patients treated at a single institution was used to identify patients with histologically confirmed ACC. Patient, tumor, and treatment characteristics were examined via review of medical records. Statistical analyses were performed to assess outcomes and associated prognostic factors. A total of 70 patients were identified with a median follow up of 3.5 years. Overall survival at 5, 10, and 15 years was 80.4%, 61.3%, and 29.4%, respectively. Disease recurrence was seen in 31.9%; of these, 72.7% developed distant metastasis. Older age, higher stage, skull base involvement, positive margins, and metastatic disease, but not local recurrence, predicted a worse overall survival. Higher stage and skull base disease were associated with shorter disease free survival. Higher T stage and perineural disease were associated with distant failure, whereas skull base involvement predicted higher local recurrence. While lung metastasis was the most common, vertebral metastasis was associated with poorer survival. In conclusion, disease stage, positive margins, skull base involvement, perineural invasion, time to recurrence, and location of metastasis, but not nodal involvement, could serve as poor prognostic factors in ACC.

INTRODUCTION

Adenoid Cystic Carcinoma (ACC) is an uncommon malignancy, accounting for 1–2% of all head and neck cancers and 10% of all salivary neoplasms.^1–4 ACC most commonly arises from the minor salivary glands and the most common site of origin is the hard palate.^2,3,5,6 The parotid is the most commonly affected major salivary gland.⁴ Peak incidence is in the fifth decade of life.^7–9

The natural history of this disease is characterized by an indolent but persistent growth rate, low likelihood of regional lymph node metastasis, high rate of distant dissemination, and eventual death.⁴ Lungs are the most common site of metastasis. Perineural invasion, even with early stage tumors, is common.^6,10 As such, ACC is one of the most aggressive salivary gland tumors. While 5 year survival rates are favorable (64–91%),^7,9,11–15 long term survival is considered poor, with 10 year survival of 37–65%.^{6,7,9,12–14,16–18} Distant metastasis occurs in 20–64%, and it is a leading cause of reduced survival in patients with ACC.^{3,15,17,19–22} Local recurrence is less common (12–40%),^16,22–24 as is cervical lymph node metastasis (6–23%).^12,16,19 Distant metastasis often occurs without local recurrence.¹⁷

Unfortunately, ACC remains a poorly understood disease because of its unpredictable nature, the need for long-term follow-up, and the scarcity of cases. In an effort to better characterize ACC, studies have attempted to identify the relationships between clinicopathological predictors including stage, clinical presentation, treatment modality and local control, distant metastases, and survival.^{6,7,10,12,13,20,22,24–27} However, there is much discordance between these studies.

The best treatment modality for ACC is also unclear.⁴ Surgical resection has historically been the mainstay of treatment. However, since the true extent of ACC tumors are often underestimated and the tumor is relatively difficult to clear surgically, positive margins occur commonly.⁶ This, along with the infiltrative nature of ACC, has prompted the frequent use of adjuvant radiotherapy. Unfortunately, conflicting evidence exists regarding this treatment approach. Studies have demonstrated reduced local recurrence with adjuvant radiation, but disagree on its impact on survival.^{1,10,11,13,25,26,28} Other analyses have shown no benefit of adjuvant radiotherapy on either disease recurrence or survival.^7,12,16,24 The objectives of our study are to report on our experience in the management of ACC with the intent of further elucidating the relevant clinicopathological prognostic factors.

PATIENTS AND METHODS

Patient Selection

Between April 1991 and February 2013, 98 patients with ACC of the head and neck underwent evaluation and treatment at the University of Wisconsin Hospital (Madison, WI). We excluded 28 patients treated only for recurrence or who had incomplete medical records. Ultimately 70 patients with a new diagnosis of ACC and primary treatment with curative intent performed at the University of Wisconsin met inclusion criteria.

Data Collection

Upon obtaining IRB approval, medical charts were reviewed to obtain data on clinical parameters, pathology, treatment, and disease outcome. TNM stages of tumors were assigned according to the staging criteria by the 2010 American Joint Committee on Cancer (AJCC), which is similar to the 2016 AJCC update. Recurrence was defined as either new metastasis or local progression after a period of disease control. Minor salivary glands were those found in the hard palate, soft palate, floor of mouth, oral tongue, base of tongue, buccal mucosa, and labial mucosa. Characteristics of resected tumors were obtained from pathology reports.

Statistical Analysis

All statistical analyses were carried out using SPSS 18.0 (Chicago, IL). Kaplan-Meier analysis was performed to determine disease free survival (DFS) and overall survival (OS). Associations between prognostic factors and survival were determined using log-rank test in univariate analyses and Cox proportional hazard model in multivariate analyses. Multivariate analysis was run on variables found to be significant on univariate analysis with the exclusion of factors that yielded a significant drop-out of our patient population. The analysis of prognostic factors associated with categorical outcomes, including the presence of local recurrence, distant metastasis, and location of metastasis, was carried out using χ² tests. A two-tailed P<0.05 was considered statistically significant.

RESULTS

Patient and Clinical Characteristics

Characteristics of the 70 study patients meeting inclusion criteria are outlined in Table 1. Median patient age was 55. Majority of patients (64.3%) presented symptomatically, with pain being the most common complaint. The parotid gland was the most common disease site. T1 and T2 stages accounted for 54.0% of disease. 8.1% of patients had nodal disease and 19.0% had skull base involvement on presentation.

Table 1:

Patient and Clinical Characteristics

Variable	Value	%	Variable	Value	%
Median Age (n=70)	55 (9–76)		T Stage (n=63)
≤55	36	51.4	T1	14	21.2
>55	34	46.6	T2	20	31.7
Sex (n=70)			T3	8	12.7
M	32	45.7	T4	21	33.3
F	38	54.3	N Stage (n=62)
Smoking (n=62)			N0	57	91.9
Never Smoked	27	43.5	N1	5	8.1
Smoking History	35	56.6	M Stage (n=65)
Symptom (n=70)			M0	63	96.9
Assymptomatic	25	35.7	M1	2	3.1
Symptomatic	45	64.3	Overall Stage (n=63)
Pain	17	24.3	1	18	28.6
Bleeding	4	5.7	2	14	22.2
Shortness of Breath	4	5.7	3	9	14.3
Airway Obstruction	3	4.3	4	22	34.9
Facial Paralysis	3	4.3	Skull Base Disease (n=63)	12	19.0
Previous Malignancy (n=70)	10	14.2	Treatment (n=70)
Head and Neck	2	20.0	Surg Alone	15	21.4
Other	8	80.0	Surg & Adjuvant Rad	41	58.6
Location (n=70)			Rad Alone	11	15.7
Major Salivary Gland	25	35.8	Chemoradiation	3	4.3
Parotid	23	32.9	Median Rad Dose (n=46)	64.8 (19.8–73.8)
Submandibular	2	2.9	Positive Margins (n=55)	17	30.9
Minor Salivary Gland	19	27.3	Invasion
Hard Palate	6	8.6	Perineural (n=64)	30	46.9
Buccal Mucosa	4	5.7	Lymphatic (n=61)	8	13.1
Labial Mucosa	2	2.9	Cartilage (n=61)	6	9.8
Floor of Mouth	2	2.9
Oral Togue	2	2.9
Base of Tongue	2	2.9
Soft Palate	1	1.4
Maxillary Sinus	7	10.0
Trachea	5	7.1
Subglottic Larynx	3	4.3
Nasal Cavity	2	2.9
Nasopharynx	2	2.9
Other	7	10.0

^*Surg = surgery, Rad = radiation

The median time between diagnosis and treatment was 14 days (0–160). Treatment distribution and pathology results are listed in Table 1. Most patients (58.6%) were treated with surgery and adjuvant radiation. Patients who received radiation alone had stage 4 disease or were not appropriate candidates for surgical resection. 30.9% of patients had positive surgical margins. Perineural, lymphatic, and cartilage invasion was seen in 46.9%, 13.1%, and 9.8% of patients, respectively.

Median and mean follow-up times were 3.5 and 5.2 years respectively (0.2–22.1). Survival and recurrence data are outlined in Table 2. Of the 31.9% patients experiencing recurrence, 72.7% experienced distant metastasis, with the lung being the most common site. Median time to recurrence was 3.3 years (0.75–10.5). The OS at 5, 10, and 15 years was 80.4%, 61.3%, and 29.4%, respectively. DFS for these time points was 65.3%, 48.0%, and 27.4%. Median survival was 6.4 years after overall disease recurrence and 3.7 years after metastasis.

Table 2:

Survival and Disease Recurrence

Variable	Value	%
Status after treatment (n=69)
Controlled	42	60.9
Persistent	5	7.2
Recurrence	22	31.9
Recurrence (n=22)
Local only	6	27.3
Distant metastasis only	11	50.0
Both	5	22.7
Location of Metastasis (n=18)
Lungs only	12	66.7
Vertebrae only	5	27.8
Both	1	5.6
Status at Last Contact (n=70)
Dead	20	28.6
Alive	50	71.4
Alive with disease	9	12.9
Alive without disease	41	58.5
Overall Survival
5yr		80.4
10yr		61.3
15yr		29.4
Disease Free Survival
5yr		65.3
10yr		48.0
15yr		47.4
Survival after recurrence
Median	6.4yrs
5yr		57.1
Survival after metastasis
Median	3.7yrs
5yr		35.9

Prognostic Factors Impacting Overall Survival

Univariate and multivariate analysis of prognostic factors impacting OS are shown in Table 3 and 4 respectively. Predictably patients over the age of 70 had a lower OS relative to patients younger than age 45 and patients with T4 disease had significantly lower survival rates compared to patients with T1 disease. Skull base involvement, positive margins, and metastatic disease resulted in significantly lower survival rates on univariate analysis, but not on multivariate analysis. While the OS at 15 years was 29.4%, none of the patients with stage 4 disease, skull base involvement, and positive margins were alive 15 years after diagnosis.

Table 3:

Univariate Analysis Predicting OS and DFS

Variable (n)	OS (%)				DFS (%)
	5yr	10yr	15yr	P	5yr	10yr	15yr	P
Age
≤55 (36)	89.6	68.7	29.3	0.062	65.7	46.8	31.2	0.634
>55 (34)	66.6	49.9	0		67.0	53.6	53.6
<45 (12)	100.0	62.5	20.8	0.011	60.6	40.4	40.4	0.767
>70 (12)	35.9	35.9	35.9		63.0	63.0	63.0
Sex
Female (38)	88.6	65.7	39.4	0.177	74.2	52.2	52.2	0.159
Male (32)	71.6	54.5	21.8		53.1	42.5	21.2
Smoker
Never (27)	85.9	61.1	61.1	0.062	71.9	53.2	35.5	0.876
Smoking Hx (35)	78.9	62.1	10.4		58.8	44.1	36.7
T stage
1&2 (34)	78.8	78.8	42	0.126	89.4	62.8	62.8	0.014
3&4 (29)	81.1	42.6	14.2		44.3	44.3	21.4
1 (14)	92.3	92.3	30.8	0.013	92.3	42.2	42.2	0.048
4 (21)	83.3	26.0	0		35.6	35.6	0
N stage
N0 (57)	79.2	63.7	29.8	0.363	70.5	53.8	41.9	0.680
N1 (5)	80.0	53.3	53.3		66.7	66.7	66.7
Overall stage
1&2 (32)	80.9	80.9	64.7	0.098	89.0	57.8	57.8	0.049
3&4 (31)	78.0	42.9	14.3		49.2	49.2	24.6
1 (18)	83.0	83.0	27.7	0.018	93.3	48.0	48.0	0.022
4 (22)	74.9	23.4	0		35.6	35.6	0
Location
Major Gland (25)	86.7	59.6	31.9	0.312	60.9	37.2	24.8	0.233
Other (45)	69.5	69.5	23.2		73.1	73.1	73.1
Skull Base
Y (12)	80.0	25.0	0	0.014	41.7	41.7	41.7	0.015
N (51)	78.8	71.7	38.4		92.8	92.8	81.2
Presentation
Asymptomatic (25)	69.3	69.3	34.6	0.575	85.0	70.8	47.2	0.200
Symptomatic (45)	83.2	48.9	12.2		52.8	35.2	26.4
Treatment
Surg (15)	92.3	92.3	92.3	0.370	80.8	80.8	80.8	0.366
Surg + Rad (41)	81.7	74.9	31.2		77.8	52.0	37.1
Radiation Dosage
≤65(23)	77.7	61.0	32.5	0.553	74.5	57.5	34.5	0.314
>65 (22)	90.5	60.0	15.0		49.5	37.1	24.7
Margins
Negative (38))	86.3	86.3	51.8	0.003	77.9	54.6	47.7	0.902
Positive (17)	77.8	51.9	0		71.4	71.4	0
Perineural Invasion
Y (30)	83.3	51.4	25.7	0.262	48.6	48.6	24.3	0.151
N (34)	92.9	82.0	32.8		81.1	48.2	48.2
Lymphatic Invasion
Y (8)	87.5	87.5	0	0.933	68.6	68.6	68.6	0.528
N (53)	80.9	65.6	30.7		69.3	45.0	32.2
Cartilage Invasion
Y (6)	80.0	80.0	0	0.528	100.0	50.0	50.0	0.539
N (53)	82.2	67.0	31.4		66.9	50.5	36.1
Metastatic Disease
Y (16)	69.9	40.8	8.2	0.002
N (53)	88.3	78.5	62.8
Local Recurrence
Y (11)	90.0	50.6	33.8	0.695
N (58)	85.3	73.2	27.9
Location of Mets
Lungs (12)	78.8	54.0	13.5	0.002
Vertebrae (5)	40.0	0	0
Time to Recurrence
<3yrs (11)	52.6	0	0	0.014
>3yrs (11)	100.0	72.7	24.2

^*OS = Overall survival, DFS = disease free survival, Y = yes, N = no, Surg = surgery, Rad = radiation

Table 4:

Multivariate Analysis Predicting OS and DFS

Variable	P	HR (95% CI)
OS
Margins	0.175	1.9 (0.8–4.8)
Metastatic Disease	0.317	2.1 (0.5–8.7)
Skull Base	0.839	0.8 (0.1–10.8)
DFS
T stage (1&2 vs 3&4)	0.503	2.1 (0.3–17.3)
Overall Stage (1&2 vs 3&4)	0.954	1.1 (0.1–9.0)
Skull Base	0.554	0.6 (0.1–3.1)

^*OS = overall survival, DFS = disease free survival, Vs = versus

Not surprisingly given their more advanced and unresectable disease, none of the 11 patients who received radiation alone were alive at 10 years. Acknowledging significant selection bias, no survival benefit was identified in patients treated with surgery and adjuvant radiation compared to surgery alone. In the event of distant failure, location of metastasis was associated with median survival (45.3 months for lung and 9.9 months for vertebrae) and patients with a shorter time to recurrence had lower survival rates (P=0.014).

Prognostic Factors Impacting Disease Free Survival

Univariate and multivariate analysis of prognostic factors impacting DFS are shown in Table 3 and 4 respectively. At 5 years, 89.4% of patients with T stage 1 and 2 disease were disease free as compared to 44.3% of patients with stage 3 and 4 disease (P=0.014). Overall prognostic stage predicted a similar pattern of DFS. Neither tumor stage nor overall stage showed significance on multivariate analysis. Skull base involvement predicted significantly shorter DFS on univariate analysis, but not on multivariate analysis.

Although 15 year DFS was lower in patients with positive margins (47.7% versus 0%), the overall difference in DFS was not significant. Patients with perineural invasion had a lower 5 year DFS (48.6% versus 81.1%), but possibly given the small number of patients without recurrence in this group, this difference was not significant.

Prognostic Factors Impacting Local Recurrence and Distant Metastasis

Table 5 summarizes the results of χ² tests analyzing the impact of prognostic factors on local recurrence and distant metastasis. Patients with higher T stage and perineural invasion experienced significantly higher rates of metastatic disease, but not local recurrence. Alternatively, skull base involvement had a significantly higher incidence of local recurrence but not distant metastasis. Patients recommended to receive adjuvant radiation experienced higher rates of distant metastasis compared to patients recommended to undergo surgery alone, most likely because of tumors characteristics that accompanied the treatment recommendations. Of note, 83.3% of patients who received surgery alone had stage 1 or 2 disease while 60.1% of patients who had surgery with radiation had stage 1 or 2 disease.

Table 5:

Chi Squared Analysis of Factors Predicting Local and Metastatic Recurrence

	Metastatic Disease		Local Recurrence
Variable	%	p	%	p
Age
≤55	36.0	0.713	24.2	0.123
>55	21.9		9.7
Sex
Female	15.8	0.075	16.7	0.900
Male	34.5		17.9
Smoker
Never	23.1	0.535	23.1	0.289
Previous/ Current	30.3		12.5
T stage
1&2	12.5	0.041	6.3	0.131
3&4	34.5		19.2
N stage
N0	23.6	0.854	11.1	0.411
N1	20.0		25.0
Overall stage
1&2	12.9	0.057	6.5	0.159
3&4	33.3		18.5
Location
Major Gland	17.4	0.368	31.0	0.214
Other	27.3		5.7
Skull Base
Y	33.3	0.410	33.3	0.033
N	22.0		8.8
Presentation
Asymptomatic	16.7	0.224	4.5	0.102
Symptomatic	30.6		20.0
Treatment
Surg	0	0.017	14.3	0.131
Surg + Rad	27.5		12.5
Rad	50.0		42.9
Radiation Dosage
≤65	34.8	0.833	13.0	0.202
>65	31.8		28.6
Margins
Negative	21.1	0.932	15.8	0.377
Positive	20.0		6.7
Perineural Invasion
Y	26.3	0.050	15.4	0.942
N	14.7		14.7
Lymphatic Invasion
Y	12.5	0.422	14.3	0.924
N	25.5		15.7
Cartilage Invasion
Y	16.7	0.668	0	0.316
N	24.5		17.0

^*Y = yes, N = no, Surg = surgery, Rad = radiation

Associations among Prognostic Factors

As metastasis to the vertebrae was found to be associated with a lower OS than lung disease, χ² analysis was performed to determine factors associated with locations of metastasis. All patients with metastasis to the vertebrae had stage T4 disease while 90% of patients with metastasis to the lung had stage T1–3 disease (P=0.018). Patients with metastasis to the vertebrae were more likely to have primary tumors located in the major salivary glands as compared to the minor salivary glands (P=0.038).

Similar analysis with other prognostic factors showed that patients with positive margins were significantly more likely to have higher stage disease (23.1%, 10.5%, 14.3%, 80.0% in T1–4 respectively, P=0.001) and higher rates of perineural involvement (54.5 versus 15.6, P=0.002). Skull base involvement was also associated with higher rates of perineural involvement (28.6% versus 6.7%, P=0.027)

DISCUSSION

ACC is a salivary gland neoplasm characterized by an indolent growth rate. Consistent with previous reports, in our series the parotid was the most common diseased major salivary gland, and the hard palate was the most common primary site of the minor salivary glands.^2–6 Our series also confirms a peak incidence in the fifth decade of life.^7–9 The majority of patients present symptomatically with pain being the most common complaint.

Our experience supports the previously reported natural history of this disease. ACC, in the majority of patients, demonstrates a slow and persistent growth rate, resulting in poor long term survival. The 5 year (80.4%), 10 year (61.3%), and 15 year (29.4%) survival rates in our series are consistent with previous reports (64–91%, 37–65%, and 25–45% respectively).^{6,7,9,11–18} The 5 year DFS rates (65.3%) seen at our institution is also comparable to recent studies (52–76%).^16,25,28 Moreover, we demonstrate a low likelihood of regional lymph node involvement (8.1%), similar to previous reports (6–10%).^12,16,19,30 Both the rate of distant metastasis and local recurrence in our series are in the lower range of other studies (20–40% and 12–40%, respectively).^{3,15–17,19–24} Our findings also demonstrate that distant metastasis commonly occurs without local recurrence and that the lung is the most common site of distant metastasis. In addition, we confirm that ACC is characterized by high rates perineural invasion, even with early stage tumors.^6,10

As expected, we find that older age reduces OS.²⁵ However, our data does not support a relationship between age and metastatic disease as previously reported by some authors.^2,19,25,26 Even these studies disagree on the nature of this association. For example, while Chen et al report that age >40 predicts a worse rate of distant control, Choi et al find that age <40 and >60 are associated with a higher rate of metastasis.

Of all the variables, tumor stage has been a consistent and reproducible prognostic variable. Our findings support the impact of both a higher overall and tumor stage on a lower OS^12,16,26,27 and DFS.^{2,10,20,27,28,31} Similar to Choi et al and Shen et al, patients with higher tumor stage disease were more likely to have metastatic disease.^11,25 However, unlike Shen et al and other studies, we did not find that tumor stage has any bearing on local recurrence.^10,11,13 In our analysis, tumor stage was associated with a higher rate of positive margins and perineural disease.

ACC involving the skull base (orbital contents/bone, sinonasal cavities, nasopharynx, cavernous sinus, clivus, pterygopalatine fossa, and infratemporal fossa) were identified by radiology reports of MRI and CT. Tumors invading the skull base are, by definition, a subset of T4 stage disease. As such, it is not unexpected that the perineural associated with T4 tumors were also common features of skull base tumors. This relationship may be secondary to the complex neural anatomy of this area. Although tumors involving the skull base are more technically challenging to resect, unlike higher stage disease, we did not find a significant association between resection margins and skull base involvement. However, in agreement with other studies, skull base disease did predict a lower OS and higher rate of local recurrence.^24,32

Although perineural invasion was seen more commonly in higher stage tumors and skull base tumors, it had no association with OS. Studies evaluating perineural disease have had conflicting outcomes, with some identifying it as an important prognostic factor^5,6,10,15,27 and others finding that it has no impact on survival.^{7,11,12,25,29} Some studies have found perineural invasion to impact local recurrence^9,26,27 and metastatic disease.^5,10,22 We find that perineural disease impacts metastatic disease but not local control.

Although nodal status on presentation has been reported to impact OS, our findings suggest that nodal involvement has no impact on OS, acknowledging small sample size, similar to Opaltek et al and Choi et al.^{9,12,25,28,29} In addition, unlike some previous reports, we did not find an association between nodal metastasis or lymphatic invasion and either local or distant disease recurrence.^12,17,25,30 Instead our data supports a hematogeneous mechanism of metastatic spread. Metastatic disease predicts a lower survival, and this is the leading cause of death after a period of disease control.^1,9,34 Given the slow progression of disease, patients with metastatic disease can remain asymptomatic for a long time. In addition, microscopic deposits to distant sites may be occurring in the early phases of the primary tumor. It is unclear how these tumors cells interact with the immune system to evade recognition.

We noted that tumors that recur more rapidly tend to be more aggressive. Patients who had recurrence within 3 years from diagnosis had a lower OS. Although the lungs are the most common site of metastasis, vertebral disease predicts a worse survival. Similar to our findings, the reported median survival after metastasis to the vertebrae (20–20.6 months) is worse than lung (32.3–54 months).^20,35 Some authors have described a lower survival with tumors originating from the minor salivary glands, and it has been suggested that this may be secondary to a higher rate of overall distant metastasis.^12,20,33,36

In contrast to some previous studies, we found that positive margins predict a lower OS.^6,16,25 Patients with a higher stage disease and perineural invasion were more likely to have positive margins. This suggests that larger tumors and those that have neural involvement may be harder to clear surgically and that obtaining negative margins is an important goal in treating ACC. In agreement with Zhang et al and Shen et al, we did not find positive margins to predict a higher disease recurrence.^2,11 This is in contrast to other studies^{6,13,22,24,26} and may be explained by positive effects of post-operative radiation, although the limitations of this study prevents a clear conclusion. However, improvements in local control with radiation may not impact survival due to the relatively high rate of distant failures seen with ACC. This could explain why many studies have shown no improvement in survival outcomes despite improved local control.^11,13,25,28

With the low rate of nodal involvement and regional metastasis seen in ACC, the need for elective lymph node dissection is debatable. Bhayani et al report occult metastasis to be as high as 23.3% in patients receiving neck dissection; and unlike our findings, studies have shown nodal status to be a predictor of disease recurrence and survival.^{9,12,19,28,29} While our data is unable to provide a strong evidence, review of the literature suggest patients who present with a clinically negative neck may have better disease control with removal of first echelon lymph nodes, although overall risk of failure in the neck is low.

As reported by Fordice et al, this study suggests the presence of two possible population of patients suffering from ACC – one with an indolent disease course and another with aggressive tumors with higher metastatic potential and lower survival rates.³⁷ Although our data suggests that higher stage and skull base diseases are more aggressive with a tendency for perineural invasion and a higher recurrence rate, the factors that drives the malignant characteristics remains unclear. Although we do not find gender to be a prognostic variable, previous reports of gender differences in recurrence rates suggest that hormonal influences, and progesterone receptor expression may partly account for the different biological behaviors in ACC.^12,18,22,40

In the future, prospective studies and multi-institutional studies with longer periods of follow-up could help address some of the inherent limitations of many single-center retrospective studies such as this one. Due to the rarity of ACC, studies tend to be small, and the relatively short duration of follow-up may also not be adequate given the protracted disease course and later recurrences seen in ACC. The percentage of recurrences that occur at 15 years after treatment is 96–97% compared to 73% at 5 years.^5,10 Our small sample size of 70 prevents meaningful multivariate analysis. Additionally, histological and molecular level investigations,³⁸ left out of this study because they were not routinely reported, may help clarify the basis of the distinct disease course. For example, solid histological pattern has been associated with worse survival and recurrence rates,^2,18,20,39 although its relationship with other prognostic variables is unclear. Finally, although we report that surgery with radiation did not reveal any significant survival benefit when compared to surgery alone, a head-to-head comparison is not possible since the study does not identify factors that may have impacted the choice of therapy, and does not eliminate selection bias related to particular form of therapy. To date, there have not been any randomized trials looking at the role of adjuvant radiotherapy and chemotherapy in patients with ACC, and this would be imperative to better guide clinical decision making.

Conclusions

ACC with higher tumor stage and skull base disease are more aggressive, have a tendency for perineural invasion, and have a higher rate of disease recurrence. Skull base disease is associated with an increased rate of local recurrence and lower survival. Alternatively, higher T stage and perineural disease are associated with metastatic failure. Distant failure, not local recurrence, impacts survival. Although lung metastasis is more common, vertebral disease predicts worse survival. Larger randomized trials will be able to better evaluate the role of adjuvant chemoradiation. Finally, further molecular level investigations are warranted to better understand the unique recurrence pattern seen in ACC.