Over 12% of stage I-III patients with EAC received surgery alone (Table 2). About 27% of patients with stage I-III EAC received tri-modality therapy (surgery, radiation and chemotherapy), while 36.5% of these patients received chemotherapy plus radiation therapy with no surgery. One-quarter of stage I-III EAC patients received no chemotherapy. The most frequently administered agent was 5-FU, frequently with cisplatin. Few patients with late/unstaged EAC received surgery, in any combination. Chemoradiotherapy, however, was given to nearly 47% of these patients. In multinomial logistic regression, age ≥ 70 was associated a 70%-80% decreased use of chemotherapy and chemoradiation in patients with EAC (Table 3).

Non-Hispanic blacks and Asian/Pacific Islanders with EAC had significantly higher hazards of cancer deaths than non-Hispanic white patients (Table 4). Patients age < 70 with a Charlson score of ≥ 1 had a significantly increased risk as did those with late/unstaged disease. EAC patients who received chemotherapy had better survival, although not statistically significant. In a separate model, EAC patients who received chemoradiation had decreased hazards (HR = 0.69, 95% CI = 0.43-1.06 model not presented). The prognostic factors in the Cox proportional hazards model containing chemoradiation were otherwise the same as those significantly associated with death in the model which adjusted for chemotherapy.

Patients with GCA received therapies at a rate between that of EAC and SAC patients. Surgery alone was provided to 34% of stage I-III GCA patients (Table 2). One-quarter of stage I-III GCA patients received trimodal therapy. Nearly twice as many GCA patients as EAC patients but less GCA than SAC patients received no chemotherapy. Fewer patients received chemoradiotherapy compared to EAC patients. 5-FU was the most frequently used chemotherapeutic agent. Nearly twice as many late/unstaged GCA patients as EAC received no therapy. In multinomial logistic regression, age ≥ 70 was associated with a 70%-80% decrease in chemotherapy alone or chemoradiotherapy (Table 3). Women and patients with a Charlson Score of ≥ 1 were significantly less likely to receive chemotherapy, but not chemoradiation. In the Cox proportional hazards models patients, with late/unstaged disease or poorly/undifferentiated tumors had an increased risk of cancer deaths while married individuals had a decreased risk (Table 4).

Of the three anatomic sites, patients with SAC were most likely to receive surgery alone (Table 2). Nearly 50% of stage I-III SAC patients received surgery alone. Less than 20% of stage I-III SAC patients received trimodal therapy. Fewer SAC patients than EAC or GCA patients received chemotherapy. As with the other two anatomic sites, 5-FU was most frequently administered. Of the three anatomic sites, late/unstaged SAC patients received no definitive cancer treatment most often. In multinomial regression, age ≥ 70 was associated with 80% less chemotherapy alone or chemoradiation (Table 3). Late/unstaged disease was associated with decreased use of chemoradiation but a substantial increased use of chemotherapy alone. Proportional hazards models for cancer deaths showed that in non-surgical patients, late/unstaged disease or a poor/undifferentiated tumor was associated with increased risk of cancer death (Table 4). However, patients receiving chemotherapy had a significantly decreased risk. Among surgical patients, a Charlson Score of ≥ 1, regardless of age and having late/unstaged disease was associated with increased hazards. Lower risks were seen among Asian/Pacific Islanders, and a non-significant decreased risk among patients who received chemotherapy (Table 4). Patients who received chemoradiation had a statistically significant decreased risk both with (HR = 0.56, 95% CI = 0.35-0.89) and without surgery (HR = 0.62, 95% CI = 0.43-0.92) (model not presented) but all other prognostic factors had similar associations with hazard ratios as the Cox models which adjusted for chemotherapy.

While there is no consensus definition of the optimal therapy for patients with resectable EAC, clinical trials have indicated survival improvements when surgery is supplemented with additional therapies. Of the three cancer sites investigated in the current study, stage I-III EAC patients had the lowest rates of surgery alone as their primary treatment, but highest rates of pre-operative chemotherapy and chemoradiation as well as definitive chemoradiation. This may reflect the significant morbidity associated with esophageal surgery[29–32]. However, toxicities associated with pre-operative chemotherapy or chemoradiation can preclude a patient from further treatment[33].

RCTs and a meta-analysis have suggested a survival benefit associated with pre-operative and adjuvant chemoradiation compared to surgery alone[1133–37]. The US-Intergroup trial, CALGB-9781, closed early due to poor accrual, but an intent-to-treat analysis on the 56 enrolled patients, demonstrated better median survival in favor of trimodal therapy[12]. In our study, over a quarter of stage I-III EAC patients received trimodality therapy. The MAGIC and FNLCC phase III trials support the use of perioperative or preoperative chemotherapy; however, we found that few (2.5%) stage I-III EAC patients received surgery and chemotherapy as primary treatment. Chemoradiation was the treatment strategy received by the largest percentage of patients with stage I-III EAC (36.5%) and late/unstaged EAC patients (47%).

Optimal therapy for GCA is not clear. Most RCTs have included patients with this cancer in trials conducted for either or both of the other two anatomic sites[1838]. Reflective of this, we found that GCA patients seemed to receive treatment at a rate that fell midway between the other two anatomic sites. In the current population-based study, stage I-III GCA patients were most frequently treated with surgery alone (34%) or trimodal therapy (22%). For late/unstaged disease less than 25% received chemotherapy alone and a significant percentage received no therapy (39%).

In contrast to EAC and GCA patients, SAC patients received surgery alone most frequently (50% of stage I-III and 16% of stage IV/unstaged disease) and radiotherapy and chemotherapy less frequently than the other two anatomic sites. Although the MAGIC and FNLCC trials demonstrated a survival advantage for patients with gastroesophageal adenocarcinoma[9], this was not evident in the current population-based study, where less than 20% of stage I-III SAC patients received chemoradiation with surgery.

With respect to advanced disease, several RCTs for SAC have demonstrated survival benefits for chemotherapy compared to best supportive care for stage IV (late-stage) disease[17]. However, we found that only 22% of patients with late/unstaged SAC received chemotherapy alone, with an additional 15% receiving chemotherapy with surgery, with surgery and radiation, or as chemoradiation. Furthermore, for late/unstaged disease, SAC patients received no therapy most frequently (43%).

Overall, our results do suggest that RCT-validated therapies have been incorporated into community practice, albeit at low levels. However, a significant percentage of patients, especially those with stage IV/unstaged disease, across all anatomic sites received no cancer-directed therapy. Our findings also highlight that the sequence and combination of chemotherapy, radiotherapy and surgery in the adjuvant setting was distinct for each anatomic site. For example, of stage I-III EAC patients who received surgery plus chemotherapy and radiation, 81% received this therapy pre-operatively [most frequently with 5-fluorouracil (5-FU) and cisplatin], while of stage I-III SAC patients who received surgery plus chemotherapy and radiation, 91% received this therapy post-operatively (most frequently with 5-FU and leucovorin). These sequences of therapy as well as the chemotherapeutic agents selected were also consistent with RCTs conducted in these disease sites[11].

Overall, the most frequently administered chemo-therapeutic agents in our study were 5-FU, cisplatin, and leucovorin. Newer agents (paclitaxel, irinotecan) have been investigated in phase II trials[3940] for use in EAC patients. We found that these drugs were used in community practice (Table 2). Use of these compounds was much lower among patients with SAC and GCA cancers. No patients received oxaliplatin, possibly because these cases were diagnosed in 2001 and findings advocating oxaliplatin for esophageal cancer were only presented in 2006[38]. Specific chemotherapeutic agents used alone or in combination with surgery and radiation are listed in Table 5. Whether patients received chemotherapy alone, chemoradiation, or trimodal therapy, the majority of patients received 5-FU in combination with another chemotherapeutic agent.

Less frequent treatment of elderly patients has been widely reported[21–2541]. Sabel et al[31] reported that 50% of patients age < 70 and 32% of those age ≥ 70 were suitable for surgery at diagnosis. Similar to this, we found that in stage I-III EAC, 30% of patients aged ≥ 70 compared to 51% of those age < 70 underwent cancer-directed surgery and 56% of gastric-cardia patients aged ≥ 70 compared to 80% of those age < 70. The age-related treatment decline is likely attributable to a number of factors: (1) Potentially higher morbidity among elderly patients; (2) Compromised treatment options due to delayed presentation by elderly patients; (3) Increased anesthesiological risk[3142]; and (4) A higher prevalence of co-morbidities.

Median age at diagnosis for stomach cancer is approximately 70 years[43], an age when patients have a reasonable life-expectancy[43]. Selected medically-fit elderly patients do as well as younger patients after surgical or adjuvant therapy[4445]. Our models indicate that in EAC or GCA patients, being age < 70 with a Charlson score of 1+ was significantly and inversely associated with surgery (data not shown). Older patients (≥ 70) were also less likely to have surgery even with a Charlson score of 0. In our multinomial models, being age ≥ 70 was associated with a decreased use of chemotherapy or chemoradiation even after adjusting for co-morbid conditions. Although we saw this disparity in the use of chemotherapy or chemoradiation by age group, there was no evidence of treatment-related differences by racial/ethnic groups. This suggests that in a community-based setting, age, in addition to co-morbidity, influences whether or not a patient receives surgery, chemotherapy, and chemoradiation.

Survival from gastric adenocarcinoma is extremely poor[46]. Theuer et al[46] reported that patients aged ≥ 70 had higher risk of death even after adjusting for clinical, non-clinical and treatment-related factors. In contrast, we observed that in GCA and non-surgical SAC patients, age and co-morbidity were not significant predictors of survival, perhaps due to the poor prognosis for these patients. We noted higher mortality in non-Hispanic blacks and Asian/Pacific Islanders with EAC. Such racial disparities have been reported in other cancers[4748], however the underlying cause for this poorer survival is not clear.

A US-based survey of 59 radiotherapy facilities indicated improved survival associated with pre-operative chemoradiation for patients with esophageal cancer[20]. In the current study, chemotherapy and chemoradiation were associated with decreased mortality for SAC and EAC patients, although not GCA patients. This suggests that the receipt of chemotherapy and/or radiotherapy may improve outcome in these poor prognosis cancers. This analysis was not a randomized study of therapy and although we adjusted for Charlson comorbidity score and additional potential confounders, patients who had better baseline health and who were selected for chemotherapy, may have been more likely to respond to such treatment or may have had better survival regardless of the use of chemotherapy.

In conclusion, RCTs have demonstrated that specific treatment strategies prolong survival in certain patient groups. We note that the use of these therapies was very low in US community-based practice despite their demonstrated survival benefits. Our study shows lower mortality among patients with EAC and SAC who received chemotherapy and significant disparities in terms of age in treatment receipt. Our findings highlight the distinctly individualized approach taken by community physicians in treating adenocarcinoma at these three anatomic sites. Community physicians appear to differentiate gastroesophageal adenocarcinoma as two distinct entities (i.e. EAC and SAC) and use different treatment strategies and chemotherapeutic agents for each, while patients with GCA are treated with a mixture of those employed for the other two anatomic sites. Improvements in community-based treatment of gastroesophageal adenocarcinoma will require better differentiation of treatments for the different anatomic sites and more extensive incorporation of those treatments proven effective in clinical trials. Future RCTs should be designed and appropriately powered to account for differences related to the anatomic site or origin of the tumor as well as the underlying tumor biology.

Randomized clinical trials (RCTs) have demonstrated that specific treatment strategies prolong survival in certain patient groups with gastric, gastro-esophageal and esophageal adenocarcinomas. However, the extent of use of these treatments in routine clinical practice is not clear. Research suggests that community-based use of treatment and the observed survival of patients in the community can vary depending on clinical and non-clinical factors.

To determine whether treatment strategies used in routine community practice are based on anatomic location or cancer origin. To examine community-based use of specific chemotherapy regimens especially those evaluated in RCTs. To assess factors that influence treatment receipt and patient survival.

We document relatively low community-based use of treatments tested in RCTs in patients with gastroesophageal adenocarcinoma. The use of these therapies was very low despite their demonstrated survival benefits. Our study shows lower mortality among patients with esophageal adenocarcinoma (EAC) and stomach adenocarcinoma (SAC) who received chemotherapy and significant disparities in terms of age in treatment receipt. Community physicians appear to take an individualized approach in treating adenocarcinoma at these three anatomic sites; differentiating between gastric and EAC and using different treatment strategies and chemotherapeutic agents for each, while patients with gastric cardia adenocarcinoma are treated with a mixture of those employed for the other two anatomic sites.

Improvements in community-based treatment of gastroesophageal adenocarcinoma will require better differentiation of treatments for the different anatomic sites and more extensive incorporation of those treatments proven effective in clinical trials. Future RCTs should be designed and appropriately powered to account for differences related to the anatomic site or origin of the tumor as well as the underlying tumor biology.

This is a retrospective study of a large number of patients with gastroesophageal adenocarcinoma focusing on treatment modalities and survival. This is an excellent and relevant study, which was well conducted and presented.