American Association of Endocrine Surgeons
Adrenocortical carcinoma: The influence of large vessel extension
Harma K. Turbendian, MD,a Vivian E. Strong, MD,b Meier Hsu, MS,” Ronald A. Ghossein, MD,d and Thomas J. Fahey III, MD,a New York, NY
Background. The impact of large vessel extension (LVE) as a prognostic factor for adrenocortical carcinoma (ACC) is not fully understood. This study aimed to assess outcome of ACC in the presence and absence of LVE.
Methods. A retrospective review of 57 patients undergoing curative intent resection for ACC over 10 years is presented comparing those with and without LVE. LVE was defined as vascular wall invasion or intraluminal extension of the neoplasm into the inferior vena cava or renal vein. Preoperative diagnostics, operative details, pathology, overall survival (OS), and recurrence-free survival (RFS) were analyzed.
Results. Multivariable regression analysis showed a significant association for decreased survival with Stage III and IV disease and LVE. Patients with LVE had more functional neoplasms, greater preoperative serum hormone levels, and more positive margins than those without LVE. Median OS was 6 years and RFS 3 years. Kaplan-Meier analysis demonstrated a significant decrease in OS and RFS with LVE. Median OS with and without LVE was 18 vs 111 months and median RFS was 11 vs 64 months. Three-year OS with and without LVE were 29% vs 93% and 3 year RFS was 15% vs 67%. Conclusion. In addition to systemic and lymph node metastases, LVE is associated with poorer OS and RFS. (Surgery 2010;148:1057-64.)
From the Department of Surgery,a New York Presbyterian Hospital-Weill Cornell; and Departments of Sur- gery, Epidemiology and Biostatistics,” and Pathology,a Memorial Sloan Kettering Cancer Center, New York, NÝ
ADRENOCORTICAL CARCINOMA (ACC) is a rare malig- nancy with an annual incidence in the U.S. of 100 to 200 cases per year. 1 Operative resection is the primary treatment modality. Delayed diagnosis, high recurrence rates, and the lack of effective adjuvant treatment account for the poor prognosis. Five-year overall survival ranges from 16% to 35%.2 Incomplete resection decreases 5-year survival by 41-62%.1,3-8 Early stage disease in the absence of lymph node and distant metastases and age younger than 40 have been shown to be positive prognostica- tors.3,5-8 Schulick and Brennan1 demonstrated median survival times of patients with stages I, II,
III, and IV disease to be 34, 40, 22, and 8 months, respectively. Pathologic features such as tumor size greater than 12 cm, abnormal mitotic figures, and intratumoral hemorrhage have been shown to be associated with poorer outcomes, and the presence of all 3 pathologic features has been shown to be fatal at uniformly 5 years follow-up.9 Stojadinovic et al10 found increased mortality in the presence of tumor necrosis, capsular and venous invasion, and adjacent organ invasion. There have been minimal data regarding the prognosis of large vessel extension (LVE) in ACC. This study aims to assess outcome in the presence and absence of LVE.
METHODS
One hundred ninety-six patients who under- went curative intent resection for ACC between 1998 and 2008 were reviewed from prospective data bases maintained at Memorial Sloan Kettering Cancer Center and New York Presbyterian Hospital - Weill Cornell. All patients with metastatic disease
Accepted for publication September 16, 2010.
Reprint requests: Thomas J. Fahey, III, MD, 525 East 68th St, New York, NY 10065. E-mail: hkt2002@med.cornell.edu.
0039-6060/$ - see front matter
doi:10.1016/j.surg.2010.09.024
at presentation had metastases resected at the initial operation or within 6 months thereof.
Demographic data, preoperative adrenal hor- mone levels, preoperative symptoms, perioperative factors, staging, time to recurrence, and time of survival were reviewed for all patients. Serum adrenal hormone levels evaluated were glucocorti- coids, mineralocorticoids, androgens, estrogens, and dehydroepiandrosterone sulfate. Functional neoplasms were defined as those that resulted in signs and symptoms of adrenal hormone excess, which included hypertension, hyperpigmentation, Cushing syndrome, masculinization, and feminini- zation. Staging was based on the most current guidelines from the American Joint Committee on Cancer.11 Stage I disease, defined as tumor less than or equal to 5 cm without local invasion, lymph node metastases, or distant metastases, was excluded from the study, because neoplasms with LVE were T3 and therefore at least stage II disease. Neoplasms greater than 5 cm without local inva- sion or metastases were classified as stage II dis- ease. Local invasion or lymph node metastases were defined as stage III disease. Stage IV disease was defined as local invasion with positive lymph nodes or distant metastases. LVE was defined as the presence of vascular wall invasion or intralumi- nal presence of neoplasm within the renal vein or inferior vena cava on either gross or microscopic examination via preoperative imaging, intraopera- tive examination, or initial pathologic assessment.
Forty-four patients underwent a clinically blinded review of tumor histology by an endocrine pathologist. Thirteen patients did not have speci- men available for review, two of which were diag- nosed with ACC based on size alone. The diagnosis of ACC was based on the 11-tiered system of pathologic features proposed by Weiss et al.12,13 Mitotic rate was determined by counting 50 high- powered fields (hpf, ×400) with an Olympus mi- croscope (U-DO model) in the areas of greatest concentrations of mitotic figures. Each 10 hpf were counted on a different slide whenever possi- ble. Tumor cell cytoplasm, diffuse architecture, tu- mor necrosis, venous or sinusoidal invasion, and capsular penetration by neoplastic cells were defined according to our previous studies on the subject.1º Microscopic resection margins were categorized as positive (neoplastic cells at the inked margin) or negative (no neoplastic cells at the inked margin). Weiss score was analyzed as a continuous variable.
A majority of patients received adjuvant chemo- therapy, radiation, or a combination of both. Because there was limited data available regarding
specific regimens and adjuvant therapy was neither randomized nor applied uniformly, adjuvant ther- apy was excluded from analysis. Recurrence was defined as clinical, pathologic, or radiologic evidence of disease.
Overall survival (OS) is defined as the time from date of resection until death or last follow-up. Recurrence-free survival (RFS) is defined as the time from date of resection until recurrence, death, or last follow-up. Two patients were ex- cluded from survival analysis due to lack of follow- up. Survival curves were estimated by the method of Kaplan and Meier,14 and comparisons were made using the log-rank test. To examine whether LVE and the reviewed clinicopathologic character- istics were associated with OS and RFS, Cox regres- sion models were used. Multivariable analyses were restricted to variables with P < . 05 on univariate analysis and with nearly complete data (includes LVE, adrenal hormone level, functional neoplasm, stage, and metastasis). Because the number of events were limited, and this limited our ability to perform traditional stepwise regression for multivariable analysis, we decided to select the best 2-variable model with the greatest score statis- tic. This selection was done separately for each endpoint. The score statistic is an overall indica- tion of the fit of the model to the observations. This method, however, does not imply that variables not in the final model were not signifi- cant predictors. Associations between LVE and clinicopathologic characteristics were assessed using Fisher exact or Wilcoxon rank sum tests. P < . 05 was considered significant.
RESULTS
Of the 196 patients reviewed, 129 patients who underwent resection for recurrences were ex- cluded. Sixty-seven patients met criteria of having undergone initial resection for histologically con- firmed ACC. Of these, 3 patients were excluded for having concurrent malignancies (lung, colon, and pancreatic adenocarcinoma). Seven patients with Stage I disease were excluded to maintain compa- rable cohorts. Of the 57 patients included in this study, 22 (39%) had LVE and 35 (61%) did not. General patient characteristics and their distribu- tion among patients with and without LVE are outlined in Table I. There were no significant differences in sex or side of neoplasm location be- tween the 2 groups. Seventy-four percent of pa- tients with LVE had increased preoperative serum adrenal hormone levels and 70% displayed symp- toms of functional neoplams as compared to
| All (N = 57) | LVE 22 (39)↑ | No LVE 35(61)* | P value | ||
|---|---|---|---|---|---|
| Sext | F | 35 (61) | 17 (77) | 18 (51) | .09 |
| M | 22 (39) | 5 (23) | 17 (49) | ||
| Age (years) | Median (range) | 48.7 (18.8-76.3) | 49.4 (21.7-73.9) | 48.0 (18.8-76.3) | .89 |
| Tumor site* | Right | 28 (49) | 11 (50) | 17 (49) | .99 |
| Left | 29 (51) | 11 (50) | 18 (51) | ||
| Increased levels of | Present | 26 (48) | 14 (74) | 12 (34) | .01 |
| adrenal hormones*+ | Absent | 28 (52) | 5 (26) | 23 (66) | |
| Functional neoplasmt | Present | 27 (49) | 14 (70) | 13 (37) | .03 |
| Absent | 28 (51) | 6 (30) | 22 (63) | ||
| Stage* | II | 24 (42) | 6 (27) | 18 (51) | .05 |
| III | 11 (19) | 3 (14) | 8 (23) | ||
| IV | 22 (39) | 13 (59) | 9 (26) | ||
| Metastases at presentation* | None | 38 (67) | 12 (55) | 26 (74) | .15 |
| Present | 19 (33) | 10 (45) | 9 (26) | ||
| EBL, ml, median (range) | 1300 (5-8000) | 2000 (250-8000) | 1100 (5-8000) | .14 | |
| Operative time, min, median (range) | 206 (70-640) | 266 (129-602) | 181 (70-640) | .01 |
*n (%).
+Missing data.
| All | LVE | No LVE | P value | ||
|---|---|---|---|---|---|
| Tumor Size, cm, median (range) | 14.5 (5.8-31.0) | 14.5 (8.0-26.0) | 14.5 (5.8-31.0) | .47 | |
| Necrosis* | Absent | 3 (7) | 0 (0) | 3 (10) | .54 |
| Present | 41 (93) | 14 (100) | 27 (90) | ||
| Mitotic rate/50 hpf* | ≤5 | 16 (32) | 4 (22) | 12 (38) | .35 |
| >5 | 34 (68) | 14 (78) | 20 (62) | ||
| Adjacent organ invasion* | Absent | 39 (89) | 11 (79) | 28 (93) | .31 |
| Present | 5 (11) | 3 (21) | 2 (7) | ||
| Margins* | Negative | 32 (68) | 9 (45) | 23 (85) | .005 |
| Positive | 15 (32) | 11 (55) | 4 (15) | ||
| Lymph nodes* | Negative | 15 (71) | 7 (64) | 8 (80) | .64 |
| Positive | 6 (29) | 4 (36) | 2 (20) | ||
| Weiss Score | Median | 7.0 (0-9) | 7.5 (5-9) | 7.0 (0-9) | .12 |
| (range) |
*n (%).
34% and 37%, respectively, in patients without LVE (P= . 01 and .03). Patients without LVE had a sig- nificantly greater proportion of stage II disease but did not have a lesser incidence of metastases at initial presentation (26% vs 45%, P = . 15). Six patients with stage II disease and LVE had renal vein extension only. Operative times were signifi- cantly greater for patients with LVE, but there was no significant difference in estimated blood loss. Pathology review (Table II) demonstrated a greater rate of positive margins in patients with LVE (55% vs 15%, P= . 005). There were no other
statistically significant differences on univariate analysis between the 2 groups.
Survival analysis was performed on 55 of 57 patients. Two patients who had no follow-up after resection were excluded. The analysis included 1 patient who died at resection. Median follow-up time among survivors was 3 years, median OS time was 6 years (95%CI, 3-15), and median RFS time was 3 years (95%CI, 1-5) (Fig 1). The median OS and RFS for patients with LVE were 17.8 months and 10.7 months respectively. These are decreases when compared to patients without LVE, whose
Surgery December 2010
A
1.0
0.9
N=55 (26 ony death)
Medion Survival: 6 Years, 95% CI (3 - 15)
0.8
Medion Follow-Up for Survivors: 3 Years
Proportion Surviving
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Years
0
2
4
6
8
10
12
14
16
18
Number at Risk:
54
32
21
12
9
5
5
4
3
0
B
1.0-
N=55 (36 Recurrence or death)
0.9
Medion RFS: 3 Years, 95% Cl: (1 - 5)
Proportion Surviving Recurrence Free
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Years
0
2
4
6
8
10
12
14
16
18
Number ot Risk:
54
27
18
10
7
4
3
2
2
0
OS and RFS were 111.4 and 63.6 months (P= . 0006 and .001). Three-year OS and RFS were also de- creased in the presence of LVE-29% vs 93% and 15% vs 67% respectively (Fig 2). Univariate analysis demonstrated that stage III or IV disease, metasta- ses at presentation, functional neoplasms, in- creased serum adrenal hormone levels, LVE, adjacent organ invasion, positive margins, and Weiss score are all associated with poorer OS and RFS (Tables III and IV). Of all variables assessed, stage and LVE were selected as the best 2-variable model based on the greatest score statistic for both OS and RFS. Stage III and Stage IV disease as well as LVE were significantly associated with OS and RFS in the selected multivariable model (Table V). Metastasis and LVE had a comparable score statistic (21.0) to stage and LVE (23.3) for OS. Furthermore, in the subgroup of patients who presented without distant metastasis, LVE was associated with both OS and RFS by univariate analysis (HR 3.26, 95%CI, 1.04-10.24, P = . 04; HR 2.49, 95%CI, 0.99-6.23, P = . 05 for OS and RFS, respectively).
A
1.0
Without LVE
0.9
With LVE
0.8
p =. 0006
Proportion Surviving
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Years
0
2
4
6
8
10
12
At risk without LVE
20
8
4
2
2
1
1
At risk with LVE
34
24
17
10
7
4
4
B
1.0
Without LVE
0.9
Proportion Surviving Recurrence Free
With LVE
0.8
0.7
P=0.001
0.6
0.5
0.4
0.3
0.2
0.1
0.0
Years
0
2
4
6
8
10
12
At risk without LVE
At risk with LVE
20
5
3
2
2
1
1
34
22
15
8
5
3
2
DISCUSSION
Complete primary resection is the only known treatment to improve survival in adrenocortical carcinoma.1,2,8,15-17 Incomplete resection is associ- ated with a 5-year survival from 0 to 9%.2 Here we report a 39% rate of LVE-which, depending on the extent of invasion, may require renal vein liga- tion, caval clamping, hepatic vascular exclusion, or cardiopulmonary bypass. To date, there have been limited data regarding the impact of LVE on OS and RFS in adrenocortical carcinoma.18 Chiche et al19 reported a series of 15 patients with a median survival time of 8 months for patients with LVE who
| Hazard ratio | 95 % CI | P value | |
|---|---|---|---|
| Lymph Node | |||
| Negative | Reference | .007 | |
| Positive | 22.08 | 2.35-207.79 | |
| Adjacent organ invasion | |||
| Absent | Reference | ||
| Present | 11.53 | 2.43-54.63 | .002 |
| Stage | |||
| Stage II | Reference | ||
| Stage III | 3.38 | 0.97-11.80 | .06 |
| Stage IV | 8.95 | 2.89-27.70 | .0001 |
| Metastases at presentation | |||
| Absent | Reference | ||
| Present | 4.18 | 1.75-10.01 | .001 |
| Functional Tumor | |||
| Absent | Reference | ||
| Present | 4.14 | 1.17-10.04 | .002 |
| Increased serum levels of adrenal hormones | |||
| Absent | Reference | ||
| Present | 4.08 | 1.67-10.00 | .002 |
| LVE | |||
| Absent | Reference | ||
| Present | 3.89 | 1.70-8.89 | .001 |
| Necrosis | |||
| Absent | Reference | ||
| Present | 3.29 | 0.42-25.79 | .26 |
| Positive margins | |||
| Absent | Reference | ||
| Present | 2.80 | 1.15-6.83 | .02 |
| Mitotic rate | |||
| <5/50hpf | Reference | ||
| >5/50hpf | 2.44 | 0.81-7.36 | .11 |
| Weiss Score | 1.44 | 1.03-1.99 | .03 |
| Tumor Side | |||
| Left | Reference | ||
| Right | 1.06 | 0.48-2.32 | .89 |
| EBL (1000 mL) | 1.03 | 0.84-1.27 | .77 |
| Age | 1.01 | 0.98-1.03 | .66 |
| Operative Time (min) | 1.00 | 0.99-1.00 | .98 |
| Tumor Size (cm) | 0.95 | 0.87-1.03 | .19 |
| Sex | |||
| Female | Reference | ||
| Male | 0.46 | 0.19-1.11 | .08 |
underwent resection. A literature review included in this report analyzed 106 patients and reported a median survival time of 9 months for patients with LVE. Fassnacht et al20 demonstrated a significantly poorer disease-specific survival for patients with tumor thrombus in either the renal vein or inferior vena cava.
Univariate analyses of OS and RFS from this series are consistent with some of the negative prognostic variables noted in several prior studies, ie, increased serum adrenal hormone levels,
symptomatic neoplasms, stage III or IV disease, metastases at presentation, adjacent organ inva- sion, positive margins, Weiss score, and positive lymph nodes. 2,4,9,12,13,21 Of these factors, we showed that patients with LVE were more likely to have increased serum adrenal hormone levels, symptomatic neoplasms, stage III or IV disease, and positive margins. There appeared to be no measurable difference in terms of histopathology or overall Weiss score except for the presence of positive margins. Our data, however, were limited
| Hazard ratio | 95 % CI | P value | |
|---|---|---|---|
| Adjacent organ invasion | |||
| Absent | Reference | ||
| Present | 13.53 | 3.53-51.94 | .0001 |
| Stage | |||
| Stage II | Reference | ||
| Stage III | 3.70 | 1.28-10.67 | .02 |
| Stage IV | 7.12 | 2.86-17.67 | <. 0001 |
| Metastases at presentation | |||
| Absent | Reference | ||
| Present | 3.50 | 1.73-7.10 | .0005 |
| Lymph Node | |||
| Negative | Reference | ||
| Positive | 3.14 | 0.81-12.14 | .10 |
| Elevated adrenal hormone levels | |||
| Absent | Reference | ||
| Present | 3.02 | 1.47-6.23 | .003 |
| LVE | |||
| Absent | Reference | ||
| Present | 2.92 | 1.48-5.74 | .002 |
| Positive margins | |||
| Absent | Reference | ||
| Present | 2.91 | 1.34-6.33 | .007 |
| Functional tumor | |||
| Absent | Reference | ||
| Present | 2.56 | 1.27-5.17 | .01 |
| Mitosis rate | |||
| ≤5/50hpf | Reference | ||
| >5/50hpf | 1.94 | 0.83-4.53 | .13 |
| Weiss score | 1.52 | 1.14-2.01 | .004 |
| EBL (mL) | 1.17 | 1.00-1.36 | .05 |
| Tumor Side | |||
| Left | Reference | ||
| Right | 1.07 | 0.55-2.09 | .84 |
| Tumor size (cm) | 1.04 | 0.98-1.11 | .18 |
| Age | 1.01 | 0.99-1.03 | .58 |
| Operative time (min) | 1.00 | 1.00-1.00 | .28 |
| Sex | |||
| Female | Reference | ||
| Male | 0.64 | 0.32-1.29 | .21 |
| Hazard Ratio for RFS (95% CL) | P value | Hazard Ratio for OS (95% CL) | P value | |
|---|---|---|---|---|
| Stage | ||||
| Stage II | Reference | <. 001 | Reference | .004 |
| Stage III | 3.81 (1.31-11.1) | 4.06 (1.13-14.6) | ||
| Stage IV | 6.24 (2.44-15.9) | 7.21 (2.24-23.2) | ||
| LVE | ||||
| Absent | Reference | .02 | Reference | .01 |
| Present | 2.32 (1.14-4.73) | 3.09 (1.27-7.54) |
by the lack of available pathology on 13 of the 57 patients. Prior studies reported a preponderance of right sided neoplasms with intracaval extension, as would be expected based on the more direct
route into the IVC via the right adrenal vein.2 We found no significant difference between the distributions of right and left sided lesions with LVE.
Metastasis is the major risk factor for survival in ACC. Our study did not show that patients with metastases at presentation were more likely statis- tically to have LVE, P= . 15 (Table I). We were able to accommodate only 2 variables at a time in our multivariate analysis. The 2-variable model with metastasis and LVE was comparable to the model with stage and LVE for OS only. Stage and LVE emerged as the best 2-variable model for both OS and RFS, as well as forward selection. Thus, LVE, controlling for stage III and IV disease (which includes all patients with metastatic dis- ease), remains significantly associated with poorer OS and RFS in our series. Nevertheless, a con- founding effect of stage on the significance of LVE cannot be completely excluded. On further examination of patients who presented without distant metastases, LVE was also significantly asso- ciated with both OS and RFS. Thus, in addition to lymph node and systemic metastases, our results suggest that LVE may be a significant predictor of poor outcome in adrenocortical carcinoma and that it may influence outcomes in ACC with the following caveat: LVE may be a process inher- ent to the natural metastatic course of ACC. Our multivariable analyses were limited by the small number of events as well as missing data (Tables I and II) in some critical factors, such as positive margins. Due to the retrospective nature of our analysis, we had little information regarding che- motherapy regimens, and thus, we were unable to account for this possible confounder. A larger study population with complete data may be able to determine whether patients with LVE in the ab- sence of known metastases truly have poorer outcomes.
Complete resection of the neoplasms and any recurrences remains the standard of care for ACC. Our results suggest that patients with ACC and LVE have a poor prognosis, making these patients candidates for adjuvant/neoadjuvant treatment protocols. Mitotane has been the only adjuvant therapy shown to have an impact on RFS. A retrospective analysis by Terzolo et al23 demonstrated a significant increase in RFS for pa- tients treated with adjuvant mitotane. A multicen- ter, randomized, prospective clinical trial-the First International Randomized trial in locally ad- vanced and Metastatic Adrenocortical Carcinoma Treatment (FIRM-ACT)- looking into the effi- cacy of adjuvant chemotherapy in ACC for pa- tients with Stage III or IV disease has completed patient accrual and is awaiting analy- sis.24 Pending the results of this trial, our data advocate for further exploration of the role of
adjuvant chemotherapy in patients with LVE as a means of improving outcomes. Studies examin- ing the mutation pattern and gene expression of ACC with LVE may provide further insight into the biology of these neoplasms and lead to better selection of candidates for trials with targeted therapies in both neoadjuvant and adjuvant settings. 25,26
This work would not have been possible without the generous contributions of Murray F. Brennan, MD, FACS, who has cared for many of these patients.
REFERENCES
1. Schulick RD, Brennan MF. Long-term survival after com- plete resection and repeat resection in patients with adre- nocortical carcinoma. Ann Surg Oncol 1999;6:719-26.
2. Schulick RD, Brennan MF. Adrenocortical carcinoma. World J Urol 1999;17:26-34.
3. Soreide JA, Brabrand K, Thoresen SO. Adrenal cortical car- cinoma in Norway, 1970-1984. World J Surg 1992;16:663-7.
4. Icard P, Chapuis Y, Andreassian B, Bernard A, Proye C. Adrenocortical carcinoma in surgically treated patients: a retrospective study on 156 cases by the French Association of Endocrine Surgery. Surgery 1992;112:972-9.
5. Zografos GC, Driscoll DL, Karakousis CP, Huben RP. Adre- nal adenocarcinoma: a review of 53 cases. J Surg Oncol 1994;55:160-4.
6. Haak HR, Hermans J, van de Velde CJ, Lentjes EG, Goslings BM, Fleuren GJ, et al. Optimal treatment of adrenocortical carcinoma with mitotane: results in a consecutive series of 96 patients. Br J Cancer 1994;69:947-51.
7. Crucitti F, Bellantone R, Ferrante A, Boscherini M, Crucitti P. The Italian Registry for Adrenal Cortical Carcinoma: analysis of a multiinstitutional series of 129 patients. The ACC Italian Registry Study Group. Surgery 1996;119:161-70.
8. Pommier RF, Brennan MF. An eleven-year experience with adrenocortical carcinoma. Surgery 1992;112:963-70.
9. Harrison LE, Gaudin PB, Brennan MF. Pathologic features of prognostic significance for adrenocortical carcinoma af- ter curative resection. Arch Surg 1999;134:181-5.
10. Stojadinovic A, Ghossein RA, Hoos A, Nissan A, Marshall D, Dudas M, et al. Adrenocortical carcinoma: clinical, morpho- logic, and molecular characterization. J Clin Oncol 2002;20: 941-50.
11. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL. AJCC cancer staging manual. 7th ed. New York: Springer; 2010.
12. Weiss LM, Medeiros LJ, Vickery AL Jr. Pathologic features of prognostic significance in adrenocortical carcinoma. Am J Surg Pathol 1989;13:202-6.
13. Weiss LM. Comparative histologic study of 43 metastasizing and nonmetastasizing adrenocortical tumors. Am Surg Pathol 1984;8:163-9.
14. Kaplan E, Meier P. Nonparametric estimation from incom- plete observations. J Am Stat Assoc 1958;53:457-81.
15. Venkatesh S, Hickey RC, Sellin RV, Fernandez JF, Samaan NA. Adrenal cortical carcinoma. Cancer 1989;64:765-9.
16. Cohn K, Gottesman L, Brennan M. Adrenocortical carci- noma. Surgery 1986;100:1170-7.
17. Icard P, Louvel A, Chapuis Y. Survival rates and prognostic factors in adrenocortical carcinoma. World J Surg 1992;16: 753-8.
18. Tritos NA, Cushing GW, Heatley G, Libertino JA. Clinical features and prognostic factors associated with adrenocorti- cal carcinoma: Lahey Clinic Medical Center experience. Am Surg 2000;66:73-9.
19. Chiche L, Dousset B, Kieffer E, Chapuis Y. Adrenocortical carcinoma extending into the inferior vena cava: presenta- tion of a 15-patient series and review of the literature. Sur- gery 2006;139:15-27.
20. Fassnacht M, Johanssen S, Quinkler M, Bucsky P, Willen- berg HS, Beuschlein F, et al. Limited prognostic value of the 2004 International Union Against Cancer staging classi- fication for adrenocortical carcinoma: proposal for a Re- vised TNM Classification. Cancer 2009;115:243-50.
21. Schteingart DE, Doherty GM, Gauger PG, Giordano TJ, Ham- mer GD, Korobkin M, et al. Management of patients with adrenal cancer: recommendations of an international consensus conference. Endocr Relat Cancer 2005;12:667-80.
22. Hedican SP, Marshall FF. Adrenocortical carcinoma with in- tracaval extension. J Urol 1997;158:2056-61.
23. Terzolo M, Angeli A, Fassnacht M, Daffara F, Tauchmanova L, Conton PA, et al. Adjuvant mitotane treatment for adrenocor- tical carcinoma. N Engl J Med 2007;356:2372-80.
24. Allolio B, Fassnacht M. Clinical review: Adrenocortical carci- noma: clinical update. J Clin Endocrinol Metab 2006;91: 2027-37.
25. Quinkler M, Hahner S, Wortmann S, Johanssen S, Adam P, Ritter C, et al. Treatment of advanced adrenocortical carci- noma with erlotinib plus gemcitabine. J Clin Endocrinol Metab 2008;93:2057-62.
26. Kerbel RS, Kamen BA. The anti-angiogenic basis of metro- nomic chemotherapy. Nat Rev Cancer 2004;4:423-36.
DISCUSSION
Dr Eren Berber (Cleveland, OH): I have a question about the statistical methods you used for multivariate analysis. Did you look at that as a yes-or-no event or did you use it as a functional time, like a Cox propor- tional hazard method? I was a little bit confused about what method you used.
Dr Harma K. Turbendian (New York, NY): We looked at that basically as a yes-or-no event. We didn’t have spe- cific data on cause-specific mortality. We didn’t have that data available.
Dr Eren Berber (Cleveland, OH): As a continuation of the Kaplan-Meier survival analysis, did you also do the multivariate analysis using a similar type of statistical method, using, for instance, the Cox proportional haz- ard ratio model, which is a very standard method to look into these oncologic type of outcomes?
Dr Harma K. Turbendian (New York, NY): Yes, it was a Cox regression model.
Dr Jeffrey Lee (Houston, TX): We certainly need all the help we can get with this rare disease and a relatively large group of patients. I’m concerned about two issues that are related to each other. One is confounding, and the second is the methods that you applied in the multivariate analysis.
With regards to confounding, stage is partially de- pendent on large vessel extension. The standard classical staging system, the Sullivan-MacFarlane system, that has been adopted now by AJCC in the 7th edition, incorpo- rates large vessel extension, and calls that a T3 tumor. And therefore, the patients are at least stage III or stage IV. Therefore, stage is partially dependent on LVE. So you have confounding right there in your multivariate analysis.
Second, you did not do, as far as I could tell, a subset analysis in which you removed the patients with distant metastatic disease. Although there wasn’t a statistically significant association with distant mets and LVE, you had about twice as many patients with LVE who have had mets, and I’m concerned about the influence of dis- tant mets on outcome confounding your findings with LVE.
Finally, it’s really not formally correct to do a multivar- iate analysis that includes both stage III and stage IV along with LVE in the same analysis. If you are going to use stage-and it’s arguable whether you should pitch it against LVE, when one depends partly on the other- you need to use stage as a single variable, stage IV versus less than IV, for example.
And I just would encourage you to redo your analyses. I think you have set things up in favor of LVE. I would encourage you to redo the analyses and let us know what those show.
Dr Richard Houghton (Boston, MA): Can you com- ment on the accuracy of preoperative imaging to predict the large vessel invasion versus no invasion?
Dr Harma K. Turbendian (New York, NY): In our study, the presence of LVE was determined either by pre- operative imaging, by findings at time of operation, or by immediate postoperative findings on pathology. Unfor- tunately, I don’t have much in terms of the accuracy of the preoperative imaging. We didn’t specifically look at that.
Dr Richard Houghton (Boston, MA): It may be some- thing worthwhile to look at, since planning is so impor- tant in these cases.
Dr Scott Wilhelm (Cleveland, OH): I was curious what imaging techniques you were using, Was this done based on MRI looking for tumor thrombus or caval invasion? Some of our Japanese colleagues, a couple of years ago at the society, presented some data on intravenous ultra- sound, looking at tumor wall invasion. Just curious if you have any information on that or not.
Dr Harma K. Turbendian (New York, NY): Any and all of the preoperative imaging was either CT scan or MRI, and I believe most of the intravascular invasion-any pa- tients that were diagnosed preoperatively-was detected on MRI, from what I recall. But there was no ultrasound involved.