Tel.: +31-30-233-8060, Fax: +31-30-233-8079
Adherence to national guidelines for gastric cancer in the Netherlands: A retrospective population-based audit
Article first published online: 30 JUL 2012
Copyright © 2012 UICC
International Journal of Cancer
Volume 132, Issue 5, pages 1156–1161, 1 March 2013
How to Cite
Ho, V. K.Y., Damhuis, R. A.M. and Hartgrink, H. H. (2013), Adherence to national guidelines for gastric cancer in the Netherlands: A retrospective population-based audit. Int. J. Cancer, 132: 1156–1161. doi: 10.1002/ijc.27718
- Issue published online: 20 DEC 2012
- Article first published online: 30 JUL 2012
- Accepted manuscript online: 9 JUL 2012 04:17AM EST
- Manuscript Accepted: 11 JUN 2012
- Manuscript Revised: 30 MAY 2012
- Manuscript Received: 22 MAR 2012
- gastric cancer;
In May 2009, a new clinical practice guideline for gastric cancer was released in the Netherlands. To determine the impact of this guideline, we evaluated trends in patterns of care, thereby focusing on the use of perioperative chemotherapy, the adequacy of lymphadenectomy and the proportion of non-curative resections. For our evaluation, we retrospectively collected information from the Netherlands Cancer Registry on 2,511 patients diagnosed with primary adenocarcinoma of the stomach during the period July 2008–June 2010, excluding tumors of the cardia. After comparing clinical management for patients diagnosed from July 2008 to June 2009 with that for patients diagnosed from July 2009 to June 2010, we conclude that our indicators for guideline adherence did not show major change, except for the proportion of patients that received an adequate lymphadenectomy (examination of ≥10 lymph nodes), which increased from 49% to 58% (p = 0.005), this increase being more pronounced for high-volume hospitals (p = 0.006). Preoperative chemotherapy was given in 45% of patients and 25% of resections was non-curative. For the total study population, the resection rate was 41% and 30-day mortality was 5.7%. However, this measure may underestimate the real operative risk for gastric cancer patients given supplementary information on postdischarge death and prolonged hospital stay.
Gastric cancer comprised about one million new cases in 2008, making it the fourth most common malignancy in the world.1 In western countries, mortality rates have shown a steady decline over the past decades, which is primarily brought about by a decrease in incidence of the disease. Five-year survival rates remain poor and vary around 23% in Europe.2 Surgical resection forms the cornerstone of curative treatment but resection rates tend to be low, mainly attributable to advanced stage at diagnosis.3 Adjuvant and neoadjuvant treatment strategies have been shown to be efficacious in the treatment of non-metastatic gastric cancer.4
In May 2009, an evidence-based clinical guideline regarding the treatment of gastric cancer was released in the Netherlands.5 An essential recommendation concerned the use of perioperative chemotherapy, consisting of three cycles of epirubicin, cisplatin and 5-fluorouracil (ECF-regimen), for patients without clinical signs of distant metastases or tumor invasion into adjacent organs. Also, patients with early-stage disease should be exempt from perioperative chemotherapy to prevent overtreatment.
Further recommendations comprised, among others, adequate lymphadenectomy with removal of at least 10 lymph nodes, and the appropriate use of palliative resections. The number of lymph nodes removed during gastrectomy is considered an important predictive factor in gastric cancer, and a more extensive removal of lymph nodes is believed to increase a patient's chance of survival. For selected patients for whom curative treatment is no longer possible, palliative surgery may be considered to improve the quality of life.
To evaluate the implementation of the new guideline in the Netherlands, we queried the Netherlands Cancer Registry (NCR) and compared patterns of care a year before and a year after publication of the guideline. We also examined resection rates and 30-day postoperative mortality during the study period and compared these with international reference figures.
The NCR is a nationwide population-based cancer registry that includes all newly diagnosed cancer cases in the Netherlands since 1989. Based on the information from the NCR, we retrospectively identified patients who were diagnosed with primary adenocarcinoma of the stomach during the time period July 2008–June 2010, thereby excluding tumors arising from the cardia (comprising the gastro-oesophageal junction, C16.0) and uncommon histotypes such as carcinoids and sarcoma.
Primary notification of diagnosed malignancies to the NCR is provided by the national registry of histo- and cytopathology (PALGA) and case ascertainment is completed through linkage with the national hospital discharge database, which contains discharge diagnoses of all patients admitted to Dutch hospitals. On notification by PALGA, specialized registrars who have access to the clinical records routinely collect information on patient status, tumor characteristics and treatments. The vital status of all patients was assessed on the 1st of February 2011 by linking with the municipal administrative databases, which register all deceased and emigrated persons in the Netherlands. Supplementary information on date of hospital discharge and laparoscopic surgery was available for patients diagnosed in 2010. The study design, data abstraction process and storage protocols were approved by the NCR review board.
Definitions and coding
Within the NCR, tumor site and histotype were registered according to the ICD-O (International Classification of Diseases for Oncology) edition of the respective period.6 Individual subsites were reassigned as fundus/corpus (C16.1,2,5,6), distal stomach (C16.3,4) and overlapping/not specified (C16.8/C16.9). The latter category also comprised patients with a primary tumor after prior gastric surgery (stump carcinoma).
Postoperative mortality was defined as death within 30 days after resection. Surgical resections were classified as partial gastrectomy, total gastrectomy or multi-organ resection. The latter definition does not include incidental splenectomy or resection of the pancreas tail. Endoscopic mucosal resections (EMR) (n = 33) were included in the analysis of resection rates but excluded for the evaluation of postoperative mortality and other performance indicators.
Due to concurrent changes of the TNM staging system, tumor stage had to be recoded according to the 6th edition of the TNM Classification of Malignant Tumors.7 Stage was assigned by combining pre-treatment (cTNM) and postsurgery (pTNM) information. To accommodate for downstaging by preoperative chemotherapy, the highest stage was selected for the analyses. For the multivariable analysis of 30-day mortality, stage IV was included as an indicator variable.
The use of neoadjuvant chemotherapy (any number of cycles) was evaluated for patients who underwent resection, excluding patients with cT1N0, cT4 or cM1 stage. Postoperative chemotherapy was evaluated for patients who underwent preoperative chemotherapy, excluding postoperative deaths and patients with residual macroscopic disease after surgery. Adequate lymph node examination was defined as ≥10 lymph nodes evaluated,8 deviating from the international standard of ≥15 lymph nodes to accommodate for a potential impact of neoadjuvant chemotherapy. Non-curative resection was assigned for patients with distant metastases and for those with microscopic (R1) or macroscopic (R2) residual disease after surgery. The intent of non-curative resection, palliative or pre-emptive, was not recorded by the NCR.
For patients diagnosed during the time period January–June 2010, supplementary information was gathered to assess additional measures of postoperative morbidity and mortality. Postdischarge death was defined as death after hospital discharge but within 30 days. In-hospital death comprised patients who died during the primary hospital stay, also if death occurred beyond 30 days after surgery. Prolonged hospital stay was assigned if patients spent >30 days in hospital during their initial stay, but were discharged alive.
Resection rates and postoperative mortality rates were tabulated by subgroups of different patient and tumor characteristics, and differences were tested for significance using chi-square tests. Prognostic factors for postoperative mortality were evaluated using multivariable logistic regression. Odds ratios are presented together with 95% likelihood ratio confidence intervals. Differences in management of patients between the two time periods of interest (July 2008–June 2009 and July 2009–June 2010) were tested for significance using chi-square tests. All tests were two-sided and p-values <0.05 were considered statistically significant. Statistical analyses were performed using STATA (version 10.0; STATA Corporation, College Station, TX).
Patient and tumor characteristics
During the study period 2,511 patients, 1,497 men (60%) and 1,014 women (40%), were diagnosed with primary adenocarcinoma of the stomach (Table 1). The median age was 73 years and 26% of patients were 80 years or older. The middle part of the stomach (fundus and corpus) comprised 27% of the tumors and 36% was located in the distal part. For the remaining 37%, the disease overlapped multiple parts of the stomach or the individual subsite could not be determined.
The resection rate was 41% overall and decreased with age from 47% for patients younger than 60 years to 29% for patients aged 80 years and older. Resection rates were similar for men and women (p = 0.59) and were considerably higher for distal tumors (51%). EMR was performed in 38 patients, followed by conventional resection in five cases.
Partial gastrectomies comprised 64% of the surgical procedures and 31% of resections were total gastrectomies (Table 2). Patients were surgically treated in 88 different hospitals and 54 hospitals performed 1–6 resections annually; 19 hospitals performed 7–9 resections and 15 hospitals performed 10 procedures or more. Of the 991 patients who underwent a surgical resection, 427 (43%) received preoperative chemotherapy. For patients younger than 70 years, preoperative chemotherapy was administered in 78% against 21% for the older group. Amongst patients who did not undergo surgery, 31% was treated by palliative chemotherapy and 3.4% by radiation therapy (data not shown).
Outcomes following surgery
The 30-day mortality was 5.7% and increased with age from 1.0% for patients younger than 60 years to 12% for patients aged 80 years and older (Table 2). Univariate analysis suggested that the postoperative risk was not related to hospital volume (p = 0.55) or extent of lymphadenectomy (p = 0.46). However, postoperative risk appeared to be higher for stage IV (8.9%) and lower after neoadjuvant chemotherapy (2.6%). Multivariable logistic regression only identified age and stage IV disease as independent prognostic factors (Table 3). For patients aged 70 years and older with stage IV disease, 30-day mortality was 11.6%. The prognostic impact of neoadjuvant chemotherapy weakened after controlling for age.
Age appeared to be an important determinant of treatment patterns (Table 4). Elderly patients were less often treated by extensive surgery, neoadjuvant chemotherapy and adequate lymphadenectomy.
For patients diagnosed in 2010, additional data were available showing that 76 patients (13%) underwent an exploratory laparoscopy or laparotomy, and postoperative mortality occurred in 18 of these patients (24%). Amongst patients that underwent a surgical resection (n = 237), laparoscopic surgery was performed in 3.8% (Table 5). Inclusion of in-hospital deaths beyond 30 days would increase the postoperative mortality estimate from 4.2% to 7.2%. No deaths were recorded after discharge, within 30 days of surgery. Prolonged hospital stay was observed in 3.8% of patients.
Trends in guideline adherence
The proportion of patients receiving preoperative chemotherapy increased from 44% in the first period to 47% in the second period (p = 0.42) (Table 6). The proportion of patients who received postoperative chemotherapy after previous preoperative chemotherapy did not change, 57% versus 56% (p = 0.87). Non-curative resections comprised 25% of all resections and this proportion was independent of age and study period. Adequate lymph node evaluation, defined as >10 lymph nodes examined, increased from 49% in the first period to 58% in the second period (p = 0.005). The improvement of adequate lymph node examination was more pronounced for high-volume hospitals (p = 0.006). Adequate lymph node evaluation was more common after neoadjuvant treatment, 61% versus 48% (data not shown).
The results from this implementation audit suggest that the introduction of the new guideline had limited impact on patterns of care. The new guideline endorses perioperative chemotherapy but that strategy was already implemented earlier, related to the publication of the MAGIC trial in 2006.9 Nonetheless, in more than half of the patients preoperative chemotherapy was not used and postoperative cycles were only given in 56% of the patients who received preoperative chemotherapy. These results corroborate the findings from a Norwegian study10 which reported that 10% of patients discontinued preoperative cycles and that only 54% of patients started postoperative cycles. Severe toxicity (≥grade 3) was observed in 34% of preoperative cycles and 50% of postoperative cycles. That finding may explain why perioperative chemotherapy is less often used in the elderly. Research should therefore focus on novel regimens that are less toxic for frail patients.
As a second performance indicator, the proportion of non-curative resections proved to be invariably high. According to the evidence table of the Dutch guideline, the merit of palliative resections is low because of the high postoperative risk and the limited survival benefit.11 Especially when two or more metastatic sites are involved, the small gain in survival time may not compensate postoperative morbidity and mortality. Nonetheless, a quarter of all resections proved to be non-curative, as defined by residual disease or positive margins. This finding may explain the 41% resection rate, which is fairly high compared with other countries. The UK Oesophago-Gastric Cancer Audit, e.g., reported a 23% resection rate for gastric cancer, probably related to the fact that in Britain almost half of the resections were preceded by staging laparoscopy.12 In contrast, a study in the United States reported a 36% resection rate for patients with stage IV disease.13 A recent review on surgery for non-curative gastric cancer included 59 articles and came to the conclusion that results are difficult to compare due to selection bias and variation in definition of treatment intent.14 To achieve a more comprehensive view of the management of gastric cancer, future audits will also need information on rates of bypass surgery and open-and-close procedures. According to the UK guideline,15 laparoscopy is required as a staging procedure for all gastric cancers and this may reduce the rates of open-and-close procedures and non-curative resections.
The Dutch guideline states that palliative resection may be considered a viable option for younger patients only. In reality, the high proportion of non-curative resections proved to be independent of age. This observation is quite significant as postoperative mortality was particularly high for patients with stage IV or high age. Moreover, the 30-day mortality rate underestimates the real operative risk, as shown by the data regarding in-hospital mortality and prolonged hospital stay. Future guidelines will require specific treatment policies for frail patients but consensus on selection criteria for resection is still lacking,15 mainly due to the absence of scientific evidence. One should, however, realize that more than a quarter of patients in this series was 80 years or older and this proportion will only increase in the coming years.
The only real progress observed during the study period concerned lymph node evaluation. Several surgical studies suggest that wider removal of lymph nodes, i.e., including N2 nodes, may improve survival under the condition that removal of pancreas or spleen is avoided.16 Pathologic assessment of >15 nodes has been suggested as a performance indicator. However, a US study reported that only 31% of community hospitals and 38% of teaching hospitals reached that standard.17 Lowering the standard to 10 lymph nodes was suggested to counteract the impact of neoadjuvant chemotherapy but our data suggest that this treatment strategy rather increases nodal clearance. In the UK, 15-nodes yield rates have increased up to 74%,12 suggesting that further progress can be made in the Netherlands, even in the high-volume hospitals.
Volume, whether related to surgeon or hospital, has become an important issue for the surgical treatment of gastric cancer. In the last 50 years, the experience with gastric surgery dramatically decreased, mainly due to the decline of peptic ulcer disease. The annual number of resections for gastric cancer simultaneously decreased due to lower incidence, better staging and a higher proportion of elderly patients. Only 15 of 88 hospitals in the Netherlands performed 10 or more resections annually. In the US, superior results were reported for hospitals performing >21 resections per year.18 A recent evaluation, however, showed that in the US 61% of gastrectomies is still performed in low-volume (<10) hospitals.19 According to a Danish audit, centralization of gastric surgery led to higher surgical quality and substantially lower in-hospital mortality.20 In contrast, the Scottish Audit of Gastric and Oesophageal Cancer reported that hospital volume did not influence long-term survival of patients undergoing gastric cancer surgery.21 In the Netherlands, surgical treatment for oesophageal cancer was formally regionalized in 2006 in hospitals performing >10 resections annually. That minimum was recently upgraded to 20. For gastric cancer, volume standards will need to be defined in a similar manner. These standards regard non-cardia gastric cancer as cardia cancer is nowadays linked to oesophageal cancer, both with respect to TNM-classification and treatment guidelines.
In conclusion, this study demonstrates that many of the guideline recommendations for gastric cancer were already implemented before the guideline was actually released. New implementation plans will need to focus on decreasing the rates of non-curative surgery. Due to the age-related morbidity of chemotherapy and surgery, future guidelines require specific attention for the management of frail patients.
- 5Oncoline. Utrecht: Comprehensive Cancer Centre the Netherlands; c2011. Dutch guideline for gastric carcinoma. Available at: http://www.oncoline.nl, accessed March 14, 2012.
- 6International classification of diseases for oncology (ICD-O), 3rd edn. Geneva: World Health Organization (WHO), 2000., , , et al.
- 7Sobin LH, Wittekind C, eds. TNM classification of malignant tumours, 6th edn. New York: Wiley-Liss, 2002.
- 12National Clinical Audit Support Programme (NCASP). Leeds: The NHS Information Centre for health and social care; c2012. National oesophago-gastric cancer audit 2010. Available at: http://www.ic.nhs.uk/services/national-clinical-audit-support-programme-ncasp/audit-reports/oesophago-gastric-cancer, accessed June 13, 2012.
- 14A systematic review of surgery for non-curative gastric cancer. Gastric Cancer, in press., , , et al.
- 17Gastric Patient Care Evaluation Group from the Commission on Cancer. Treatment of gastric adenocarcinoma may differ among hospital types in the United States, a report from the National Cancer Data Base. J Gastrointest Surg 2007; 11: 410–9; discussion 419–20., , , et al.;
- 20Enforcing centralization for gastric cancer in Denmark. Eur J Surg Oncol 2010; 36( Suppl 1): S50–4., , , et al.