Ann Surg Oncol (2014) 21:306–314 DOI 10.1245/s10434-013-3303-0

ORIGINAL ARTICLE – THORACIC ONCOLOGY

Factors Associated with Local–Regional Failure After Definitive Chemoradiation for Locally Advanced Esophageal Cancer Arya Amini, MD1,2, Jaffer Ajani, MD3, Ritsuko Komaki, MD1, Pamela K. Allen, PhD1, Bruce D. Minsky, MD1, Mariela Blum, MD3, Lianchun Xiao, MS4, Akihiro Suzuki, MD3, Wayne Hofstetter, MD5, Stephen Swisher, MD5, Daniel Gomez, MD1, Zhongxing Liao, MD1, Jeffrey H. Lee, MD6, Manoop S. Bhutani, MD6, and James W. Welsh, MD1 Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX; 2UC Irvine School of Medicine, Irvine, CA; 3Department of Gastrointestinal Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX; 4Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, TX; 5Department of Thoracic and Cardiovascular Surgery, The University of Texas MD Anderson Cancer Center, Houston, TX; 6Department of Gastroenterology, Hepatology and Nutrition, The University of Texas MD Anderson Cancer Center, Houston, TX 1

ABSTRACT Background. Locally advanced esophageal cancer is often treated with a trimodality approach. While a substantial proportion of such patients initially achieve a clinical complete response (cCR) after chemoradiation, only a small proportion achieve durable control. We analyzed patients who reached cCR after definitive chemoradiation for esophageal cancer to identify clinical predictors of local disease recurrence. Methods. We identified 141 patients who obtained initial cCR after definitive chemoradiation without surgery for esophageal cancer from 2002 through 2009. The initial response to treatment was assessed by endoscopic evaluation and biopsy results, with cCR defined as having no evidence of disease present. Patterns of failure were categorized as in-field (within the planned treatment volume [PTV]), outside the radiation treatment field, or both. Results. At a median follow-up of 22 months (range, 6– 87 months), 77 patients (55 %) had experienced disease recurrence (local or both). Of first failures, 32 (23 %) were

Electronic supplementary material The online version of this article (doi:10.1245/s10434-013-3303-0) contains supplementary material, which is available to authorized users. Ó Society of Surgical Oncology 2013 First Received: 21 November 2012; Published Online: 7 November 2013 J. W. Welsh, MD e-mail: [email protected]

outside the radiation field, followed by 30 (21 %) within the field, and 15 (11 %) were both. By multivariate analysis, in-field failure after cCR was associated with a pretreatment standardized uptake value on positron emission tomography of [10 (subhazard ratio [SHR] 3.31, p = 0.023) and poorly differentiated tumors (SHR 3.69, p = 0.031). All failures, in-field and out-of-field, correlated with non-Caucasian ethnicity (SHR 2.55, p = 0.001), N1 disease (SHR 2.05, p = 0.034), T3/T4 disease (SHR 3.56, p = 0.011), and older age (SHR 0.96, p = 0.008). Conclusions. Our data suggest that selected clinical characteristics can be used to predict failure patterns after definitive chemoradiation. Such risk-assessment strategies can help individualize therapy. Esophageal cancer is the eighth most common cause of cancer and the sixth most common cause of death from cancer, with 482,000 new cases and 407,000 deaths estimated worldwide in 2008. Localized esophageal carcinomas are highly aggressive and difficult to cure, as they often persist or recur after definitive chemoradiation.1 Although surgery continues to be the standard approach for most locally advanced esophageal cancers, cure rates after surgery alone have been poor, with 3- to 5-year survival rates ranging from 6 to 35 %.2–4 However, the morbidity associated with surgery after initial chemoradiation can be considerable, with perioperative mortality rates reported as high as 10–20 % in some centers. Individuals with esophageal cancer are often older, heavy smokers, and malnourished secondary to disease.5

Predictors in Esophageal Cancer Response

The benefit of trimodality therapy for this purpose may not outweigh the risk of serious morbidity for all patients. Two previous European randomized studies comparing primary chemoradiation versus chemoradiation followed by surgery reported improvement in local control but no difference in overall survival.6,7 Both trials also noted high rates of treatment-related mortality (9–14 %). The trial reported by Stahl et al. found local progression-free survival rates to be higher among those who underwent surgery, but treatmentrelated mortality rates were also higher after surgery (12.8 vs. 3.5 % for those who did not have surgery, p \ 0.0001).6 The RTOG (RTOG 0246) recently completed a trial reporting that outcome of patients who achieved cCR after chemoradiotherapy who were observed was similar to those who received chemoradiotherapy followed by surgery.8 With a median follow-up of 22 months, the 1-year survival was 71 %, and it demonstrated that a selective surgical approach is possible, yet the trial did not meet its primary endpoint of increased overall survival. These findings suggest that perhaps the subset of patients who respond favorably to neoadjuvant chemoradiotherapy may benefit from a selective approach to therapy, utilizing surgery only in patients with documented residual or recurrent local/regional disease. The advent of improved imaging and biopsy techniques such as endoscopic ultrasound with fine needle aspiration has improved the accuracy of identifying disease status assess depth of infiltration and nodal status after neoadjuvant chemoradiation before surgical resection.9 More accurate restaging raises the question of whether all patients who achieve a clinical complete response (cCR); that is, those whose biopsy specimens of primary tumor and lymph nodes with the support of imaging show no evidence of disease after chemoradiation require surgery. Several studies have shown that such patients have significantly improved disease control and overall survival relative to those without a complete response at the time of surgery.10 This suggests that some patients may not need immediate surgery to achieve successful cure and, furthermore, may fare better with observation and salvage surgery if necessary.11,12 To test this hypothesis, we assessed a variety of patient- and disease-related characteristics for potential associations with patterns of failure, especially local failures that would otherwise be removed by esophagectomy. Our reasoning was that such factors would be useful in predicting risk, personalizing therapeutic approaches, and perhaps identifying patients who otherwise would not need surgery. Hence we retrospectively analyzed patients who had had an initial cCR after definitive chemoradiation and identified clinical predictors of disease recurrence, both within and outside the original radiation treatment field.

307

PATIENTS AND METHODS We retrospectively identified 141 patients who had had cCR after definitive chemoradiation for locally advanced esophageal cancer at The University of Texas MD Anderson Cancer Center from January 2002 through January 2009. Local failures (defined as those within the original radiation field or planning treatment volume [PTV]) and out-of-field failures were identified by comparing posttreatment positron emission tomography (PET), computed tomography (CT), or PET/CT scans with the original CT-based radiation treatment plans. Baseline tumor stage was assessed according to the American Joint Committee on Cancer criteria for esophageal carcinoma, sixth edition. The institutional review board of MD Anderson Cancer Center approved this post hoc analysis. Treatment Simulation, Planning, and Delivery For treatment simulation and planning purposes, most patients had undergone 4-dimensional (4D) CT scanning to account for respiratory motion. During acquisition of the 4D CT images, patient respiration was monitored with an external real-time position management respiratory gating system (Varian Medical Systems, Palo Alto, CA, USA). From the 4D-CT acquisition, a maximum-intensity projection image set was created to help define the internal gross tumor volume. The gross tumor volume (GTV) was contoured on the planning CT scans by the attending radiation oncologist using all available resources, including data from PET/CT fusion scans, endoscopic ultrasonography images, and diagnostic CT images. The GTV was expanded to the clinical target volume (CTV) by extending the radiation coverage 3 cm superiorly, 3 cm inferiorly, 1 cm laterally, and 3 cm into the gastric mucosa. The PTV was then generated by using a uniform 0.5-cm expansion beyond the borders of the CTV. All organs at risk (heart, lung, and liver) were outlined. All patients were to be treated to a total dose of 50.4 Gy, given in 28 fractions, with concurrent fluorouracil plus an additional agent, most commonly cisplatin, docetaxel, or paclitaxel. All radiation was to be delivered as intensitymodulated radiation therapy; treatment plans were generated with the Pinnacle planning system (Phillips Medical Systems, Andover, MA, USA). Some patients received induction chemotherapy before chemoradiation, and others received concurrent chemoradiation alone. Chemotherapy agents used during induction therapy were predominantly fluorouracil, cisplatin, paclitaxel, and docetaxel. The choice of chemotherapy was based on physician preference. All patients included in the study had biopsy-proven negative disease after definitive chemoradiation and

308

A. Amini et al.

TABLE 1 Patient characteristics Characteristic

Value or no. of patients (%)

Sex Female Male Age; years, median (range)

23 (16 %) 118 (84 %) 68 (43–90)

Clinical tumor status T1

7 (5 %)

T2

16 (11 %)

T3

109 (77 %)

T4

8 (6 %)

TX

1 (1 %)

Clinical lymph node status N0 N1

49 (35 %) 92 (65 %)

Clinical metastatic status M0

117 (83 %)

M1a

15 (11 %)

M1b

9 (6 %)

Statistical Analysis

Primary tumor site Cervical Upper thoracic Mid-thoracic Lower thoracic

2 (1 %) 14 (10 %) 15 (11 %) 110 (78 %)

Tumor histology Adenocarcinoma

107 (76 %)

Squamous cell

33 (23 %)

Neuroendocrine

1 (1 %)

Tumor grade G2 (moderately diff.) G3 (poorly diff.) Tumor length; cm, median (range)

65 (46 %) 76 (54 %) 5 (0–17)

Induction chemotherapy No

88 (62 %)

Yes

53 (38 %)

Concurrent chemotherapy No Yes

necessarily cCR unless proven by biopsy. Pathologic specimens were obtained by upper endoscopy with ultrasoundguided biopsy, and all specimens were analyzed at MD Anderson. Pretreatment and posttreatment characteristics and follow-up disease recurrences were then extracted from the medical records of these individuals. Treatment failures were identified from serial posttreatment images that included PET scans (using SUV maximum), CT scans, and endoscopic images.11 Failures were confirmed by biopsy during endoscopy. Failure location (within or outside the PTV) was identified by fusing current PET scans with the treatment plan CT scan. For simplicity, any failure within the radiation treatment volume was considered local (because these volumes encompassed prophylactic nodal coverage), and any failures outside the radiation treatment volume were considered out-of-field. Those that first failed both in and out-of-field were categorized as both. Failures were both pathologically proven and documented by serial radiography.

2 (1 %) 139 (99 %)

Radiation dose; Gy, median (range)

50.4 (39–66)

Pretreatment SUVmax, median (range)

10.3 (0–42)

Posttreatment SUVmax, median (range)

3.8 (0–13)

SUV change, median (range)

6.3 (0–40)

diff differentiated, SUV standardized uptake value

required a minimum of 1 follow-up upper endoscopy study to confirm this. No patients received additional therapy (adjuvant chemotherapy or surgery) between completion of initial treatment and disease relapse. Biopsy-negative disease on endoscopy was the only criteria necessary to establish cCR.13 Patients with negative PET scans were not

Data analysis was performed using Stata/MP 11 statistical software. Pearson’s Chi square was used to assess measures of association in frequency tables. The equality of means for continuous variables was assessed using the t test. A p value of 0.05 or less was considered to be statistically significant, and all statistical tests were based on a 2-sided significance level. Competing risk-regression analysis using the Fine and Gray method for in-field failure or death, out-of-field failure or death, or any failure (diseasefree survival) or death was applied.14 Multivariate models were assessed using backward elimination with all variables having a p value of 0.25 or less included in the assessment. When in-field failures were assessed, out-offield failure (DM failure status) was included in the assessment for both univariate and multivariate analysis, and the same was done for out-of-field assessment. Multivariate models were assessed with and without these outcome variables (LR failure and DM failure). The pretreatment SUV cutoff of 10 and posttreatment SUV (from posttreatment PET scan) max cutoff of 3.5 were used in the multivariate analysis was chosen as a dichotomous variable, as this represents the value just below the median value of 10.3 and 3.8, respectively. RESULTS Patient Characteristics Table 1 shows the baseline characteristics of the 141 patients included in this retrospective analysis, with a

Predictors in Esophageal Cancer Response

309

a

b Site of first failure

Proportion 1.00

In-field failure

0.75

In-field 21% No failure 45%

0.50 Out-of-field 23%

0.25

Both 11%

0

12

24

Number at risk 132 93 46

c

36

48

60

72

84

96

11

2

2

0

60

72

84

96

2

2

0

Time (months) 29

15

d

Proportion 1.00

Out-of-field failure

Proportion 1.00

0.75

0.75

0.50

0.50

0.25

0.25

0

12

24

Number at risk 132 93 46

36

48

60

72

84

96

Time (months) 29

15

11

2

2

0

0

12

All failures

24

Number at risk 132 93 46

36

48

Time (months) 29

15

11

FIG. 1 Pie graph representing overall sites of first failure (a) and Kaplan–Meier curves demonstrating relapse-free survival (RFS) for first site in-field failures (b), first site out-of-field failures (c), and any disease recurrence over time (d)

median follow-up of 22 months (range, 6–87 months). Of all living patients, median follow-up was 36 months (range, 11–85 months). The median patient age was 68 years (range, 43–90 years); the majority (118, or 84 %) were male. All patients had clinical staging (AJCC sixth edition); most patients had T3 tumors (77 %), N1 disease (65 %), M0 disease (83 %), and poorly differentiated tumors (54 %). The most common tumor histology was adenocarcinoma (76 %), and most tumors (78 %) were located in the lower thoracic esophagus and included types I, II, and III (gastroesophageal junction). Median tumor length was 5 cm (range, 0–17 cm), and the median baseline standardized uptake value (SUV) on PET was 10.3 (range, 0–42) (PET values were obtained at time of first follow-up after completion of chemoradiation). Nearly all patients (99 %) received concurrent chemotherapy, and 38 % received induction chemotherapy as well. The median radiation dose was 50.4 Gy (range, 39–66 Gy).

Posttreatment PET scan was performed at a median of 46 days (range, 22–181 days) after treatment. Posttreatment EGD with biopsy confirming cCR was performed at a median of 85 days (range, 37–219 days). Reasons for no surgery included patient choice for observation (32.6 %), poor performance status before chemoradiation (25.5 %), poor performance status after treatment (2.8 %), and initial presence of unresectable disease (31.2 %). Locoregional free survival at 2 and 5 years was 80 and 64 %, respectively. The 2- and 5-year overall survival was 59 and 32 %, respectively, for our study population. Patterns of Failure Of the entire study population, 77 (55 %) had disease recurrence, either in-field, out-of-field, or both. First sites of failure were in-field in 30 cases (21 %), out-of-field in 32 (23 %), and both in 15 (11 %); 64 patients (45 %) did

310

A. Amini et al.

TABLE 2 Univariate competing risk analysis of local (in-field) failures Variable

Hazard ratio (95 % CI)

TABLE 2 continued Variable [3.5

Sex Male

1.93 (0.43–8.72)

0.392

1.00 (0.97–1.04)

0.813

(continuous variable)

0.112

1.16 (1.03–1.31)

0.018

C6.3

0.75 (0.34–1.69)

0.494

1.00 (0.96–1.03)

0.804

1.86 (0.84–4.10)

0.125

0.41 (0.15–1.12)

0.083

\6.3 0.79 (0.19–3.18)

0.737

Caucasian History of alcohol consumption

(continuous variable) Tumor length C5 cm

Frequenta

1.35 (0.49–3.72)

0.557

Occasional

1.17 (0.33–4.09)

0.806

\5 cm Out-of-field failures

None

Yes

Smoking status

No

Current

0.64 (0.16–2.58)

0.531

Everb

1.34 (0.44–4.06)

0.603

Never

95 % CI 95 %confidence interval, diff differentiated, SUV standardized uptake value a

Alcohol status: frequent (C1 drink/day), occasional (\7 drinks/ week)

Baseline body mass index [30

1.92 (0.86–4.30)

SUV change

Ethnicity Non-Caucasian

p value

B3.5

Female Age (continuous variable)

Hazard ratio (95 % CI)

p value

0.44 (0.15–1.25)

0.122

–

–

0.92 (0.41–2.05)

0.841

1.23 (0.46–3.31)

0.683

Middle

1.55 (0.35–6.90)

0.563

Lower

0.85 (0.26–2.83)

0.796

1.99 (0.87–4.55)

0.102

0.85 (0.32–2.27)

0.747

1.11 (0.51–2.41)

0.801

0.34 (0.04–2.60)

0.296

2.25 (0.94–5.38)

0.068

1.02 (0.99–1.05)

0.162

b

Smoking status: ever (quit [1 year)

B30 Tumor status T3/T4 T1/T2 Nodal status N1 N0 Metastatic status M1 M0 Tumor site

Cervical/upper Tumor grade G3 (poorly diff.) G2 (moderately diff.) Tumor histology Squamous

not have disease recurrence (Fig. 1a). Of note, time between the last day of treatment and follow-up endoscopy proven cCR was not a significant predictor for in-field failure (p = 0.75), out-of-field failure (p = 0.28), any disease relapse (p = 0.79), or overall survival (p = 0.31). Of first site in-field recurrences, 77 % (23 of 30) were at the primary site, 17 % (5 of 30) were at a local lymph node, and 7 % (3 of 30) recurred at the primary and nodal site. For first site out-of-field failures, 75 % (24 of 32) were organ metastasis, 16 % (5 of 32) involved lymph nodes, and 9 % (3 of 32) failed first at both an organ and lymph node. Overall relapse-free survival (RFS) for first site infield and out-of-field recurrences, in addition to overall all failures at 5 years was 57.9, 66.7, and 32.1 %, respectively (Fig. 1b–d). Predictors of In-Field Failure After Clinical Complete Response

Adenocarcinoma Induction chemotherapy Yes No Radiation dose [50.4 Gy B50.4 Gy Pretreatment SUV [10 B10 (continuous variable) Posttreatment SUV

By univariate analysis, no statistically significant predictors of locoregional (in-field) failure were found. There was a trend for those with a pretreatment SUV [10 to have increased in-field failures (p = 0.068) (Table 2). By multivariate analysis, several factors were associated with in-field failure. Pretreatment SUV[10 had an in-field failure rate of 31.3 % compared with 20.7 % with SUV B10 (subhazard ratio [SHR], 3.31; 95 % confidence interval [95 % CI], 1.18– 9.30; p = 0.023). Poorly differentiated (grade III) were also at higher risk of failure, with 26.3 % compared with 15.4 % in-field recurrences with moderately differentiated (grade II)

Predictors in Esophageal Cancer Response

311

TABLE 3 Multivariate competing risk analysis for local (in-field) failures Variable

Hazard ratio (95 % CI)

p value

0.46 (0.07–3.24)

0.437

1.00 (0.92–1.09)

0.917

0.90 (0.21–3.91)

0.885

–

–

1.24 (0.30–5.07)

0.763

0.74 (0.13–4.35)

0.743

Predictors of Out-of-Field Failure After Clinical Complete Response

3.69 (1.13–12.08)

0.031

0.57 (0.13–2.49)

0.459

1.04 (0.26–4.09)

0.958

By univariate analysis, predictors for out-of-field failures include younger age (p = 0.007), N1 disease (p = 0.001), poorly differentiated tumors (p = 0.005), tumor length C5 cm (p = 0.029), radiation dose [50.4 Gy (p = 0.035), and having a posttreatment SUV [3.5 (p = 0.035) (Supplementary Table 1). Predictors for outof-field failures by multivariate analysis included N1 disease (SHR, 5.89; 95 % CI 1.30–26.66; p = 0.021) (Supplementary Table 2).

0.28 (0.02–5.06)

0.391

Sex Male Female Age (continuous variable) Ethnicity Non-Caucasian Caucasian Tumor status T3/T4 T1/T2 Nodal status N1

Among those with pretreatment SUV [10, site of first failure included 24 % in-field, 30 % out-of-field, 15 % both in-field and out-of-field, and 31 % did not recur. For poorly differentiated tumors, 26 % recurred in-field, 26 % out-of-field, 12 % both in-field and out-of-field, and 36 % did not recur. For advanced stage patients (stage CII), site of first failure were similar compared with all stages combined; 22 % in-field, 24 % out-of-field, 10 % both infield and out-of-field, and 44 % did not recur. For all stage patients, in-field failure patterns based on T stage were 0 % (0 of 5) for T1, 12.5 % (2 of 16) for T2, 24.0 % (26 of 109) for T3, and 12.5 % (1 of 8) for T4. For T3N0 and T3N1, respectively, in-field failure rates were 30.3 % (10 of 33) and 21.1 % (16 of 76). Lastly, in regard to tumor histology, in-field failure rates were 12.1 % (4 of 33) and 24.3 % (26 of 107) for squamous cell and adenocarcinoma accordingly.

N0 Metastatic status M1 M0 Tumor grade G3 (poorly diff.) G2 (moderately diff.) Tumor histology Squamous Adenocarcinoma Induction chemotherapy Yes No Radiation dose [50.4 Gy

Predictors of Disease-Free Survival After Clinical Complete Response

B50.4 Gy Pretreatment SUV [10

3.31 (1.18–9.30)

0.023

2.01 (0.74–5.46)

0.173

1.40 (0.39–5.00)

0.602

0.30 (0.08–1.12)

0.073

B10 Posttreatment SUV [3.5 B3.5 Tumor length C5 cm \5 cm Out-of-field failures Yes No 95 % CI 95 %confidence interval, diff differentiated, SUV standardized uptake value

(SHR, 3.69; 95 % CI, 1.13–12.08; p = 0.031) (Table 3). The 5-year RFS was worse for pretreatment SUV[10 (47.0 vs. 63.7 %) and poorly differentiated tumors (44.0 vs. 71.8 %) (Fig. 2).

In univariate analysis for all failures, both in-field and out-of-field, after cCR, an increased risk of disease recurrence was found to be statistically significant for nonCaucasians (p \ 0.001), younger patients (p = 0.003), frequent alcohol consumption (p = 0.006), T3/T4 tumors (p = 0.002), N1 nodal status (p = 0.004), poorly differentiated tumors (p = 0.005), those presenting with a pretreatment SUV[10 (p = 0.011) and posttreatment SUV [3.5 (p = 0.006), and tumor length C5 cm (p = 0.009) (Supplementary Table 3). In multivariate analysis for all failures, an increased risk of disease recurrence was again found in non-Caucasian compared with Caucasian individuals (SHR, 2.55; 95 % CI 1.50–4.32; p = 0.001), T3/T4 tumors (SHR, 3.56; 95 % CI 1.34–9.47; p = 0.011), and N1 versus N0 disease (SHR, 2.05; 95 % CI 1.06–3.99; p = 0.034). Protective factors in multivariate analysis included older age (SHR, 0.96; 95 % CI 0.93–0.99; p = 0.008) and squamous cell histology

312

A. Amini et al.

a

b

Pre-treatment SUV Log rank P = 0.2350

Proportion

1.00

0.75

0.75

0.50

0.50

0.25

Baseline SUV 10 Baseline SUV > 10 0

12

24

36

48

60

72

84

96

Log rank P = 0.0265

Proportion

1.00

0.25

Tumor histology

Baseline SUV 10 Baseline SUV > 10 0

12

24

Time (months) Number at risk 54 41 25 54 38 15

14 10

7 6

5 4

36

48

60

72

84

96

1 1

1 1

0 0

Time (months) 1 0

1 0

0 0

Number at risk 61 46 27 71 47 19

18 11

9 6

7 4

FIG. 2 Kaplan–Meier curves demonstrating relapse-free survival (RFS) for all patients with in-field failure based on having a pretreatment SUV [10 (a) and poorly differentiated (grade III) tumors (b)

(SHR, 0.42; 95 % CI 0.19–0.94; p = 0.035) (Supplementary Table 4). DISCUSSION The use of preoperative chemoradiation for esophageal cancer has increased substantially at MD Anderson and elsewhere, and nearly one-third of patients have no evidence of viable tumor at the time of surgery, after neoadjuvant chemoradiation.15 However, for patients who achieve a cCR we have no reproducible way of predicting which ones benefit from further surgery and which should be observed. We identified several pretreatment risk factors associated with in-field failures including pretreatment SUV value greater than 10 and poorly differentiated tumors. These factors may help to personalize treatment by identifying patients most likely to benefit from a selective surgical approach. The selective approach is now gaining acceptance given the favorable outcomes demonstrated from RTOG 0246.8 This is an especially important consideration for older patients or those with other comorbid conditions that would make esophagectomy more risky. In addition, a recent study analyzing 266 patients from a prospective database who were treated with neoadjuvant chemoradiation compared outcomes between those receiving surgery within 8 weeks of neoadjuvant treatment versus after 8 weeks.16 The results showed that the timing of surgery was not associated with perioperative complications, pathologic response, or overall survival, suggesting that a delay in surgery does not necessarily lead to worse outcomes.

For patients undergoing esophagectomy, age is a significant risk factor for pulmonary complications and death after esophagectomy. Respiratory complications are twice as likely, and mortality 4 times as likely, in patients older than 70 years.5 Our study went further to identify additional factors associated with any recurrence type (both infield and out-of-field), which included non-Caucasian ethnicity, T3/T4 tumors, N1 nodal status, and adenocarcinoma histology. Several of these predictive factors have been identified as prognostic indicators in prior studies of patients with complete response to neoadjuvant chemoradiation as well, although the patients in those trials received surgery in addition to preoperative therapy. We demonstrated that a pretreatment SUV [10 correlated significantly higher with in-field failures compared with those with SUV B10. As SUV correlates with tumor activity and size, we would expect greater SUV to be more resistant to local treatment, given the greater tumor burden present, whereas N1 nodal status, for example, better defines risk for distant failure as it signifies lymph node spread that can then invade other neighboring organs. There are several studies that demonstrate SUV to be a predictor for esophageal cancer recurrence, making it a useful prognostic test.17,18 T stage has also been a welldocumented prognostic indicator for disease recurrence. For example, studies have suggested patients with earlystage disease (T1 or T2 tumors) may not benefit from additional surgery, as such tumors are less likely to fail both locally or distantly compared with more advanced T3 and T4 tumors after neoadjuvant chemoradiation.19 We also found that patients who were older were less likely to

Predictors in Esophageal Cancer Response

experience recurrent disease, either in-field or out-of-field. Although this finding may be confounded by death from other causes, it raises the question of whether older patients benefit from surgery after chemoradiation. One study in which patients with stage I or II esophageal cancer who were age C65 years were identified in the surveillance, epidemiology, and end results-medicare database found no significant difference in overall survival in patients with squamous cell cancer who underwent chemoradiation or esophagectomy, suggesting that in elderly patients, chemoradiation may be sufficient.20 With the use of more effective chemotherapy regimens in addition to improved radiation delivery techniques, we expect that the number of patients with cCR after neoadjuvant chemoradiation will increase. While induction chemotherapy did not reduce distant failure in our study, likely confounded as those who received induction chemotherapy had more advanced disease, studies have suggested response to induction chemotherapy can improve prognosis.6 In addition, proton therapy to doses of 70–90 Gy (RBE) has been shown in several studies to produce high complete clinical response rates, as high as 78 %.21,22 Although RTOG 94-05 revealed no benefit for radiation dose escalation, it was performed in the pre-IMRT era.23 We have previously demonstrated that with advanced IMRT treatment planning we could selectively increase the dose to the GTV by 28 % while also reducing the dose to both the heart and lung compared with traditional 3D techniques.24 Similar benefits can also be achieved from the dosimetric advantage of intensity modulated protons therapy.25 This study did have a number of limitations. These include its retrospective nature, with the associated biases, and the fact that techniques for disease staging improved considerably over the 7-year span of the study. Moreover, although all patients had pathologically confirmed lack of primary or nodal disease, some of the patients who later experienced in-field failure may have had false-negative biopsy findings. Also the median follow-up was only 22 months, and more in-field failures are likely to develop over time. Lastly, because of the low number of in-field failures (28 of 141), our findings will need to be validated in other, perhaps larger data sets in the future. Some of the strengths of our analysis include the large number of patients, including those with cCR, the consistency of the treatment dose and technique used over the period of study, and the relatively long follow-up time. In summary, we identified several risk factors for infield recurrence in patients who had a cCR after definitive chemoradiation. Our findings suggest that among patients with cCR, those who initially presented with a pretreatment SUV [10 and poorly differentiated tumors are at greater risk for in-field recurrences and should be prospectively studied to see if they could avoid immediate surgery,

313

delaying esophagectomy until the time of proven in-field recurrence. This strategy may be particularly relevant for patients who are elderly or have comorbid conditions that would make surgery difficult to tolerate. Future studies are needed to identify further clinical as well as biologic predictors to identify patients who are likely to achieve cCR after neoadjuvant chemoradiation and to identify, among those patients, those who may not need surgical therapy after the chemoradiation. ACKNOWLEDGMENT Research supported in part by donations from the Family of M. Adan Hamed, Dallas, Park, Sultan, and Smith families; the Carlos H. Cantu Family foundation; the Rivercreek and Schecter Private Foundations; the Kevin M. and Debra L. Frazier Foundation. DISCLOSURES interest.

The authors declare no potential conflicts of

REFERENCES 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J Clin. 2011;61:69–90. 2. Kelsen D, Winter KA, Gunderson LL, Mortimer J, Estes NC, Haller DG, et al. Long-term results of RTOG trial 8911 (USA Intergroup 113): a random assignment trial comparison of chemotherapy followed by surgery compared with surgery alone for esophageal cancer. J Clin Oncol. 2007;25:3719–25. 3. Bosset JF, Gignoux M, Triboulet JP, Tiret E, Mantion G, Elias D et al. Chemoradiotherapy followed by surgery compared with surgery alone in squamous-cell cancer of the esophagus. N Engl J Med. 1997;337:161–7. 4. Walsh TN, Noonan N, Hollywood D, Kelly A, Keeling N, Hennessy TP. A comparison of multimodal therapy and surgery for esophageal adenocarcinoma. N Engl J Med. 1996;335:462–7. 5. Welsh J, Settle SH, Amini A, Xiao L, Suzuki A, Hayashi Y, et al. Failure patterns in patients with esophageal cancer treated with definitive chemoradiation. Cancer. 2012;118:2632–40. 6. Stahl M, Stuschke M, Lehmann N, Meyer HJ, Walz MK, Seeber S, et al. Chemoradiation with and without surgery in patients with locally advanced squamous cell carcinoma of the esophagus. J Clin Oncol. 2005;23:2310–7. 7. Bedenne L, Michel P, Bouche´ O, Milan C, Mariette C, Conroy T, et al. Chemoradiation followed by surgery compared with chemoradiation alone in squamous cancer of the esophagus: FFCD 9102. J Clin Oncol. 2007;25:1160–8. 8. Swisher SG, Winter KA, Komaki RU, Ajani JA, Wu TT, Hofstetter WL, et al. A phase II study of a paclitaxel-based chemoradiation regimen with selective surgical salvage for resectable locoregionally advanced esophageal cancer: initial reporting of RTOG 0246. Int J Radiat Oncol Biol Phys. 2012;82:1967–72. 9. Bohanes P, Yang D, Chhibar RS, Labonte MJ, Winder T, Ning Y, et al. Influence of sex on the survival of patients with esophageal cancer. J Clin Oncol. 2012;30:2265–72. 10. van Hagen P, Hulshof MC, van Lanschot JJ, Steyerberg EW, van Berge Henegouwen MI, Wijnhoven BP, et al. Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med. 2012;366:2074–84. 11. Visser EP, Boerman OC, Oyen WK. SUV: from silly useless value to smart uptake value. J Nucl Med. 2010;51:173–5.

314 12. Berger AC, Farma J, Scott WJ, Freedman G, Weiner L, Cheng JD, et al. Complete response to neoadjuvant chemoradiotherapy in esophageal carcinoma is associated with significantly improved survival. J Clin Oncol. 2005;23:4330–7. 13. Taketa T, Correa AM, Suzuki A, Blum MA, Chien P, Lee JH et al. Outcome of trimodality-eligible esophagogastric cancer patients who declined surgery after preoperative chemoradiation. Oncology. 2012;83:300–4. 14. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1994;236:376–84. 15. Hofstetter W, Swisher SG, Correa AM, Hess K, Putnam JB Jr, Ajani JA, et al. Treatment outcomes of resected esophageal cancer. Ann Surg. 2002;236:376–84; discussion 384–5. 16. Kim JY, Correa AM, Vaporciyan AA, Roth JA, Mehran RJ, Walsh GL, et al. Does the timing of esophagectomy after chemoradiation affect outcome? Ann Thorac Surg. 2012;93:207–12; discussion 212–3. 17. Suzuki A, Xiao L, Hayashi Y, Macapinlac HA, Welsh J, Lin SH, et al. Prognostic significance of baseline positron emission tomography and importance of clinical complete response in patients with esophageal or gastroesophageal junction cancer treated with definitive chemoradiotherapy. Cancer. 2011;117:4823–33. 18. Suzuki A, Xiao L, Hayashi Y, Blum MA, Welsh JW, Lin SH, et al. Nomograms for prognostication of outcome in patients with esophageal and gastroesophageal carcinoma undergoing definitive chemoradiotherapy. Oncology. 2012;82:108–13. 19. Rohatgi P, Swisher SG, Correa AM, Wu TT, Liao Z, Komaki R, et al. Characterization of pathologic complete response after

A. Amini et al.

20.

21.

22.

23.

24.

25.

preoperative chemoradiotherapy in carcinoma of the esophagus and outcome after pathologic complete response. Cancer. 2005;104:2365–72. Abrams JA, Buono DL, Strauss J, McBride RB, Hershman DL, Neugut AI. Esophagectomy compared with chemoradiation for early stage esophageal cancer in the elderly. Cancer. 2009; 115:4924–33. Mizumoto M, Sugahara S, Nakayama H, Hashii H, Nakahara A, Terashima H, et al. Clinical results of proton-beam therapy for locoregionally advanced esophageal cancer. Strahlenther Onkol. 2010;186:482–8. Lin SH, Komaki R, Liao Z, Wei C, Myles B, Guo X, et al. Proton beam therapy and concurrent chemotherapy for esophageal cancer. Int J Radiat Oncol Biol Phys. 2012;83:345–51. Minsky BD, Pajak TF, Ginsberg RJ, Pisansky TM, Martenson J, Komaki R, et al. INT 0123 (Radiation Therapy Oncology Group 94-05) phase III trial of combined-modality therapy for esophageal cancer: high-dose versus standard-dose radiation therapy. J Clin Oncol. 2002;20:1167–74. Welsh J, Palmer MB, Ajani JA, Liao Z, Swisher SG, Hofstetter WL, et al. Esophageal cancer dose escalation using a simultaneous integrated boost technique. Int J Radiat Oncol Biol Phys. 2012;82:468–74. Welsh J, Gomez D, Palmer MB, Riley BA, Mayankkumar AV, Komaki R, et al. Intensity modulated proton therapy allows dose escalation and normal-tissue sparing in locally advanced distal esophageal tumors. Int J Radiat Oncol Biol Phys. 2011;81:1336– 42.