Tumor differentiation impacts response to neoadjuvant therapy and survival in patients with esophageal adenocarcinoma

Abstract

Objective:

Tumor differentiation and stage have been known since the 1940s to be useful to prognosticate outcomes in patients with gastrointestinal adenocarcinoma. For esophageal adenocarcinoma the current staging system only considers tumor grade in early T1–2N0 tumors where poor differentiation is a known risk factor for failure of endoscopic therapy and lymph node metastases. The relevance of tumor differentiation in patients with locally advanced tumors undergoing neoadjuvant therapy is less understood. The aim of this study was to evaluate the impact of tumor differentiation on response to neoadjuvant therapy and survival after esophagectomy for adenocarcinoma.

Methods:

A multi-institution retrospective chart review was undertaken of all esophageal cancer patients who underwent neoadjuvant therapy followed by esophagectomy from 1/2010–12/2017. Patients with tumors that were well differentiated (G1) or moderately differentiated (G2) were combined and compared to those with poorly differentiated tumors (G3).

Results:

There were 550 patients with esophageal adenocarcinoma, 485 men (88.2%) and 65 women. The median age was 61 years, and the tumor was G1/2 in 288 patients (52.4%) and G3 in 262 patients. Clinical stage prior to neoadjuvant therapy was similar for patients with G1/2 vs G3 lesions and was III or higher in 77% with 66% of patients being node positive. Post-resection pathology showed pathologic complete response (pCR) in 87 patients (15.8%). The frequency of pCR vs residual disease was similar in patients with G1/2 tumors and those with G3 lesions, but those with G3 tumors were significantly more likely to have residual disease both in the esophagus and in the lymph nodes. Median follow-up was 63 months and absolute survival, overall survival and disease-free survival were all significantly worse in patients with G3 tumors. Further, even with pCR, patients with G3 tumors had significantly worse survival compared to those with G1/2 tumors.

Conclusions:

This study showed that poor differentiation in patients with local-regionally advanced esophageal adenocarcinoma is associated with worse survival compared to patients with G1/2 tumors, even among those with pCR after neoadjuvant therapy and esophagectomy. These results suggest patients with poorly differentiated tumors should be considered for adjuvant therapy even with pCR after esophagectomy.

INTRODUCTION

Poor tumor differentiation in esophageal adenocarcinoma is a marker for a more aggressive tumor and worse prognosis. The 8^th edition of the American Joint Commission on Cancer (AJCC) staging for esophageal cancer included tumor grade (Gx: can’t be assessed; G1: well differentiated; G2: moderately differentiated; and G3 poorly differentiated) in the staging categories. However, for adenocarcinoma of the esophagus tumor grade is only used in the clinical staging of T1–2 N0 tumors.[1] Currently most patients present with local-regionally advanced disease with clinical T3 tumors for which tumor grade is not taken into account. Data on the importance of tumor differentiation primarily comes from patients with superficial esophageal adenocarcinoma where it is recognized that poor differentiation is a risk factor for failed endotherapy and portends an increased risk for lymph node metastases.[2–4] Consequently, tumor grade is used to determine therapy in patients with superficial adenocarcinoma of the esophagus. The relevance of tumor grade in patients with local-regionally advanced adenocarcinoma treated with neoadjuvant therapy and esophagectomy is not well defined. The aim of this study was to evaluate the impact of tumor differentiation on the response to neoadjuvant therapy and survival after esophagectomy in patients with local-regionally advanced adenocarcinoma.

METHODS

A multi-institution retrospective chart review was undertaken of all esophageal cancer patients who underwent neoadjuvant therapy followed by esophagectomy from January 2010 to December 2017. Deidentified data was collected into a study database. The contributing institutions were MD Anderson Cancer Center, Memorial Sloan Kettering Cancer Center, The Oregon Clinic/ Providence Portland Cancer Center, Swedish Cancer Institute, the University of Rochester, the University of Southern California, and the University of Toronto. Local IRBs at each participating center including the IRB at Portland Providence Medical approved this study.

All patients included in the study had adenocarcinoma of the esophagus treated with neoadjuvant therapy followed within 12 weeks by an esophagectomy. The esophagectomy in all patients was with an Ivor Lewis or McKeown approach that included an abdominal and thoracic lymphadenectomy. Patients were excluded for squamous cell histology, a synchronous or other malignancy within 5 years of the esophageal cancer diagnosis, chemoradiotherapy inconsistent with NCCN guidelines, a transhiatal esophagectomy, salvage esophagectomy, a non-R0 resection, or recurrent disease or death within 90 days of the esophagectomy. All patients were restaged in accordance with the AJCC Tumor-Nodal-Metastasis 8^th edition. Patients that had adjuvant therapy after resection were included in the analysis.

For the analyses, patients with tumor grade 1 (well-differentiated) and 2 (moderately differentiated) were combined (G1/2) and compared to those who had tumor grade 3 (poorly differentiated) lesions (G3). This was done because of the small number of patients that had tumor grade 1 lesions and evidence in prior studies that tumor grades 1 and 2 are biologically different from tumor grade 3.[2, 3] Residual disease after esophagectomy on final pathology was defined as no residual disease found in the specimen or in any resected lymph nodes (pathologic complete response, pCR), residual disease only within the esophagus, residual disease only in the lymph nodes, or residual disease in the esophagus and lymph nodes. Recurrent disease was considered local if it occurred within the residual esophagus or the graft, regional if in lymph nodes within the area of resection including the mediastinum and upper abdomen, and systemic if it occurred in distant lymph nodes and / or organs.

Statistical analysis

The statistical analysis was performed using the statistical package R version 4.2.3. Descriptive statistics were presented as mean, standard deviation, number, and percentage. Categorical variables were summarized as number and percentages and compared using Chi-square test or Fisher exact test. Continuous variables were summarized as medians and interquartile ranges (IQRs) and compared using the Student’s t or Wilcoxon rank-sum test whenever appropriate. The primary outcome measures were time in months from surgical procedure to death due to any cause for overall survival (OS) and time in months from surgical procedure to disease recurrence for disease-free survival (DFS). Median survival is given with 95% confidence intervals (CI). The Kaplan-Meier method was used for survival analyses for OS and DFS according to pre-treatment tumor grade and pathologic response to neoadjuvant therapy (pCR versus residual disease). Univariate analysis was performed to assess differences in survival with respect to tumor grade and response to neoadjuvant therapy using the log rank test. Comparisons were considered statistically significant at a two-tailed p ≤ 0.05.

RESULTS

There were 550 patients with esophageal adenocarcinoma and complete follow-up included in this study, 485 men (88.2%) and 65 women. The median age at the time of esophagectomy was 61 years (range 22–83). The tumor was G1/G2 in 288 patients (52.4%) and G3 in 262 patients. The pre-therapy clinical stage was III or higher in 77% of patients, and 66% were lymph node positive, with no significant difference in clinical stage between patients with G1/2 vs G3 tumors (Table 1). Post-resection pathology showed pCR in 87 patients (15.8%). The frequency of pCR versus residual disease was similar in patients with G1/2 and G3 tumors. However, the location of residual disease differed between groups. Patients with G1/2 tumors were significantly more likely to have residual disease in the esophagus only, while patients with G3 tumors were significantly more likely to have residual disease in both the esophagus and lymph nodes (Table 2). The median follow-up for survivial was 63 months (95% CI 64.4–72.6) with no significant difference based on tumor grade. The overall frequency of recurrent disease was 31%, and was 27.1% for patients with G1/2 tumors versus 34.4% for patients with G3 tumors, but this did not reach statistical significance. The type of recurrence was similar in both groups. Among the 254 deaths (44.5%), the majority (85%) where the cause was known were due to cancer.

Table 1.

Patient and tumor characteristics

	Grade 1/2 n=288	Grade 3 n=262	p value
MEDIAN AGE	61 years (22–82)	61 years (33–83)
SEX	35 (12.2%) females	30 (11.5%) females	0.79
	253 (87.8%) males	232 (88.5%) males
MEDIAN BMI	27 (18–46)	27 (14–49)
SURGERY TYPE
IVOR LEWIS	247 (85.8%)	209 (79.8%)	0.06
MCKEOWN	41 (14.2%)	53 (20.2%)
CLINICAL STAGE
T1–2	45	17	*0.0021
T3	237	240
T4	4	3
UNKNOWN	2	2
N1	97	85	0.78
N+	190	175
N1	183	168
N2	4	7
N3	3	0
UNKNOWN	1	2
M0	288	258	0.14
M1	-	2
UNKNOWN	-	2
I	4	1	0.17
II	63	49
III	214	202
IV	6	8
UNKNOWN	1	2

Table 2.

Postneoadjuvant therapy and esophagectomy outcomes

	Grade 1/2 n=288	Grade 3 n=262	p value
MEDIAN FOLLOW-UP SURVIVING PATIENTS	64.5 months (8.4–130.6)	61.8 months (1.4–131.3)	0.094
pCR	44 (15.3%)	43 (16.4%)	0.72
RESIDUAL DISEASE	244 (84.7%)	219 (83.6%)	0.72
ESOPHAGUS ONLY	140 (57.3%)	92 (42%)	*0.0009
NODES ONLY	9 (3.7%)	10 (4.6%)	0.63
BOTH	95 (38.9%)	117 (53.4%)	*0.0017
ADJUVANT THERAPY	31 (10.8%)	30 (11.5%)	0.79
RECURRENT DISEASE	78 (27.1%)	90 (34.4%)	0.065
LOCAL	11 (14.1%)	12 (13.3%)	0.88
REGIONAL	15 (19.2%)	12 (13.3%)	0.29
DISTANT	50 (64.1%)	62 (68.9%)	0.51
UNKNOWN	2 (2.6%)	4 (4.4%)	0.51

Median overall survival was 80.3 months (95% CI 65.4-NA). The survival at 5 and 10 years was 57% (95% CI 53%−62%) and 44% (95% CI 39%−50%) respectively. Overall survival is shown in Figure 1, and was significantly worse in patients with G3 tumors (p=0.001). Survival in patients with pCR based on tumor grade is shown in Figure 2, and was also significantly worse in patients with G3 tumors (p=0.003). Similar to overall survival, patients with G3 tumors had worse disease-free survival (p=0.003) [Figure 3] and worse disease-free survival after pCR compared to patients with G1/2 tumors (p=0.008) [Figure 4]. As expected, patients with pCR had better disease-free survival than those with residual disease (p< 0.001) [Figure 5]. Absolute survival at 5 years was 55.2% in patients with G1/2 tumors and 40.2% in those with G3 tumors (p=0.0021).

Figure 1:

OS in patients by tumor grade (G1/2 versus G3)

Figure 2:

OS in patients with pCR after neoadjuvant therapy by tumor grade (G1/2 versus G3)

Figure 3:

DFS in patients by tumor grade (G1/2 versus G3)

Figure 4:

DFS in patients with pCR after neoadjuvant therapy by tumor grade (G1/2 versus G3)

Figure 5:

DFS based on pCR versus residual disease

Discussion:

In 1890 David Paul von Hansemann, a professor of pathological anatomy at the University of Berlin, proposed that cancer cells were characterized by loss of differentiation and asymmetrical mitoses.[5] His ideas were highly disputed since at the time Rudolf Virchow and other prominent pathology professors believed that cancer cells were cytologically no different from normal cells, only their behavior differed.[6] Finally in 1920 it was shown that tumors of the same histologic type can behave differently based on the degree of differentiation. In his landmark paper Albert Broders, a surgical pathologist, divided patients with squamous cell cancer of the lip into 4 histologic groups based on the degree of differentiation and showed that this allowed him to predict their prognosis.[7] This paper started the era of personalized cancer care, which now a hundred years later is mainstream.[6] In the early 1930s Cuthbert Dukes was able to replicate the findings of Broders in patients with rectal adenocarcinomas. He also noted that poorly differentiated tumors grew more quickly and were less likely to be surgically curable, and from this he introduced the concept of clinical tumor staging based on the depth of tumor invasion and presence of lymph node metastases.[8] Subsequently, in 1940 Broders published another landmark study demonstrating how tumor grade based on differentiation and staging were complimentary for prognostication, and both are accepted now in the management of patients with gastrointestinal malignancies.[9]

Currently, in patients with esophageal adenocarcinoma it is well established that poor tumor differentiation is associated with an increased risk for lymph node metastases.[3, 4] However, in in the 8^th edition of the AJCC staging manual, tumor grade is used only in the clinical staging of T1–2N0 adenocarcinomas.[1] In patients with superficial esophageal adenocarcinoma undergoing endoscopic therapy, poor differentiation has been associated with an increased risk for failure and disease recurrence.[2] When there is early invasion into the submucosa, poor differentiation is one of the factors that makes the lesion high risk for lymph node metastases and influences the decision about endoscopic therapy versus esophagectomy in these patients.[10] However, although tumor grade impacts the clinical stage for T3 N0 squamous esophageal cancer in the upper or mid esophagus, it does not alter the clinical stage for a T3 adenocarcinoma.

In this study we sought to determine whether tumor grade would impact the results of neoadjuvant therapy and survival after esophagectomy in patients with local-regionally advanced adenocarcinoma, typically T3 lesions with and without lymph node metastases. Using data from multiple centers across the US and Canada we showed that compared to G1/2 lesions, G3 lesions had a worse absolute, overall and disease-free survival despite similar clinical staging prior to neoadjuvant therapy. In part, this may reflect the deficiencies in clinical staging of esophageal cancer. Although the “T” stage was higher in patients with G3 tumors, the “N” stage and overall stage was similar to those with G1/2 lesions. One would expect patients with G3 lesions to be more likely to have node metastases, and perhaps they did since the frequency of residual disease in both the esophagus and lymph nodes was significantly higher in G3 patients on final resection pathology. Interestingly, the frequency of pCR was similar between groups, suggesting that tumor grade did not significantly impact the response to neoadjuvant therapy. However, overall and disease-free survival in those with pCR was significantly worse with a G3 lesion. This finding would suggest that despite achieving pCR with neoadjuvant therapy, patients with G3 tumors are at increased risk for death from cancer during follow-up and may benefit from adjuvant therapy to a greater extent than those with G1/2 tumors. Since only about 10% of patients in our study received adjuvant therapy, we are unable to assess the impact of adjuvant therapy on survival. However, this will be the focus of future studies in this disease.

In patients with residual disease, the residual disease was more likely to be limited to only the esophagus in patients with G1/2 tumors. In a prior study we found this to be associated with improved survival compared to residual disease in the lymph nodes only or the esophagus and lymph nodes (publication pending). In contrast, patients with G3 tumors were significantly more likely to have residual disease in the esophagus and lymph nodes. This finding helps explain the worse overall and disease-free survival in these patients. The impact of residual disease on survival in patients with esophageal adenocarcinoma is well established in the literature, and was confirmed in this study where disease-free survival was significantly worse in patients with residual disease compared to those with pCR.

Although survival was impacted in patients with G3 lesions both in those with pCR and with residual disease after neoadjuvant therapy, this modern data with an oncologic sound esophagectomy that included both thoracic and abdominal lymphadenectomy led to a median overall survival of 80.3 months and 5 and 10-year survival of 57% and 44% respectively. In patients with G1/2 lesions that had pCR after neoadjuvant therapy the survival at 5 years exceeded 75%. Unfortunately, pCR was seen in only approximately 16% of patients, underscoring the continued importance of esophagectomy and lymph node dissection in these patients. It remains to be seen if watchful waiting and salvage esophagectomy can lead to similar survivial as we show in this study with a traditional approach of planned esophagectomy following completion of the neoadjuvant therapy. The survival in this study stands out when compared to older studies in patients with esophageal adenocarcinoma, once largely considered a death sentence. Likely the new era of adjuvant immunotherapy in patients at increased risk for recurrent disease will continue to improve the likelihood of long-term survival in these patients.

This study is strengthened by the multi-institutional collaboration to allow greater numbers of patients, and the requirement for an esophagectomy that included a systematic thoracic and abdominal lymphadenectomy. However, it is weakened in that the different centers did not use the same type of neoadjuvant therapy and follow-up protocol. Further, as in most cancer studies determining cause of death, cancer vs non-cancer, is fraught with difficulty, and in our patients the cause of death was unclear in 56% of those that died during follow-up. However, in those with a known cause of death it was cancer-related in 85%, and since both the overall and disease-free survival curves show the same findings, it is unlikely that knowing the cause of death in all patients would have substantively changed our findings. Lastly, our dataset did not include biomarkers and future studies should explore how tumor grade and biomarkers interact for treatment and survival.

In conclusion, tumor grade has been widely accepted to impact the risk of lymph node metastases and survival in patients with gastrointestinal malignancies. Currently it is used in the management of patients with superficial esophageal adenocarcinoma but not in those with local-regionally advanced disease who are treated with neoadjuvant therapy prior to esophagectomy. This study showed that although pre-neoadjuvant therapy clinical stage was similar in patients with G1/2 versus G3 tumors, overall and disease-free survival was significantly worse with a G3 tumor. Further, although tumor grade did not impact the frequency of pCR with neoadjuvant therapy, survival after pCR was significantly better in patients with G1/2 tumors. Inclusion of tumor grade into the staging system for local-regionally advanced esophageal adenocarcinoma may improve prognostication of survival and contribute to further individualization of cancer treatment.