Cholangiocarcinoma in Denmark: time trends in mortality following resection

Abstract

Background:

Cholangiocarcinomas have a poor prognosis, and resection is the only curative-intent treatment. We examined time trends in mortality risk after resection in a nationwide Danish cohort.

Methods:

We identified all Danish citizens diagnosed with cholangiocarcinoma in 2004–2022 who subsequently underwent liver resection, biliary tract resection, or a combination of the two. We computed the proportions of patients who received downstaging or adjuvant chemotherapy, and we estimated age- and sex-adjusted mortality risks.

Results:

Between 2004 and 2022, 611 patients with cholangiocarcinoma underwent resection (13% of 4572 patients with cholangiocarcinoma). The proportion who underwent resection increased from 9% in 2004–2012 to 16% in 2018–2022, the median age of resected patients rose from 65 to 69 years, and the proportion with comorbidities increased from 17% to 28%. In patients who underwent liver resection, the proportion who had received downstaging chemotherapy increased from <7% in 2004–2012 to 12% (8–20), and the proportion who received adjuvant chemotherapy increased from 27% (18–39) in 2004–2012 to 49% (40–58) in 2018–2022. Despite the increasing use of chemotherapy before and after liver resection, age- and sex-adjusted mortality remained essentially the same, with 1-year mortality ranging from 14% to 20% (P-value = 0.37). In patients who underwent biliary tract resection, only 3% received downstaging chemotherapy, but the proportion who received adjuvant therapy increased from 17% (10–29) in 2004–2012 to 59% (48–68) in 2018–2022. Even so, these patients’ age- and sex-adjusted mortality remained stable (1-year mortality ranging from 19% to 24%, P-value = 0.34).

Conclusion:

The proportion of patients with cholangiocarcinoma who underwent liver or biliary tract resection increased, and more patients received downstaging chemotherapy and adjuvant chemotherapy. Nonetheless, age- and sex-adjusted mortality risks remained constant. The explanations for this pattern are unclear but likely involve changing patient characteristics other than age and sex.

HIGHLIGHTS

• Cholangiocarcinomas have a poor prognosis, but resections are performed with curative intent.

• The proportion that underwent resection increased between 2004 and 2022, as did the proportion that received downstaging or adjuvant chemotherapy.

• The median age and the prevalence of comorbidity increased among patients who underwent resection for cholangiocarcinoma between 2004 and 2022.

• The mortality risk following resection was stable between 2013–2017 and 2018–2022.

Introduction

Cholangiocarcinomas have a poor prognosis, and resection is the only curative-intent treatment. However, the prognosis remains bleak even after resection[1]. Disconcertingly, the incidence of cholangiocarcinoma has increased globally^[2,3], and it has also increased in Denmark, driven by an increase in intrahepatic cholangiocarcinoma[4]. Fortunately, numerous advances in surgical procedures, oncological treatments, and imaging techniques have been made in recent decades, which may have reduced patient mortality.

In the last 20 years, only one study has reported on time trends in mortality risk following resection in patients with cholangiocarcinoma. This study from the Netherlands reported that between 2009 and 2016 mortality risk after a resection was stable in patients with distal cholangiocarcinoma[5], but the study did not report absolute mortality risks, and it ended before the current standard-of-care adjuvant chemotherapy was introduced.

We need up-to-date studies of current trends to understand how newer treatments might have altered the prognosis for patients. This information is important not only for surgeons and their patients, but also for researchers who design trials of future treatments. Therefore, we aimed to describe time trends in mortality risk after resection of cholangiocarcinoma in a Danish nationwide cohort.

This cohort study has been reported in line with the STROCSS guidelines[6].

Material and methods

We conducted an observational cohort study using routinely collected Danish healthcare data and identified all Danish patients diagnosed with cholangiocarcinoma between 2004 and 2022 who underwent resection within 1 year of diagnosis (to allow for diagnostic workup and downstaging chemotherapy). We computed the proportion of patients who underwent resection of the liver and/or the extrahepatic bile ducts [including pancreaticoduodenectomies (Whipple’s procedure)] and mortality risks from the time of resection. All data were analyzed using “Stata Statistical Software: Release 18.5” and “R Statistical Software: v4.3.3.”

Data sources

Data were obtained from the nationwide healthcare registries in Denmark. We identified patients through the National Patient Registry and the Cancer Registry: The National Patient Registry contains data on surgical procedures [coded according to the NOMESCO Classification of Surgical Procedures (NCSP)], non-surgical procedures including administration of chemotherapy, and in- and outpatient contacts with discharge diagnosis codes [(International Classification of Diseases, 10th revision) ICD-10 codes since 1994][7]. The Cancer Registry contains data on all cancers diagnosed in Danish patients since 1943[8]. The Civil Registration System assigns a unique personal identification number to all Danish residents at birth or upon immigration, and it can be used to link individual-level data across the Danish healthcare registries. Furthermore, data on date of birth, death, and migration are contained in the Civil Registration System[9]. This study did not require written consent from patients or approval from an ethics committee, according to Danish law.

Study cohort

First, we identified all Danish adults (≥18 years) who received an ICD-10 diagnosis code for cholangiocarcinoma (C22.1, C24.0, C24.8, and C24.9, including subcodes) between 1 January 2004 and 31 December 2022 (N = 4572). We identified patients from both the National Patient Registry and the Cancer Registry, as we have previously demonstrated similar validity of cholangiocarcinoma diagnosis codes in both registries[4]. Second, we included all 611 patients who underwent resection of the liver or extrahepatic bile ducts (including Whipple’s procedure) within 1 year after their first cholangiocarcinoma diagnosis [NCSP codes, resection of liver: JJB, resection of extrahepatic bile ducts: JKC10-40, JKC96-97, and JLC30-40 (Whipple’s procedure), all including subcodes]. Liver transplantation was not included, as liver transplantation for cholangiocarcinoma is very rare in Denmark.

The codes we used to define chemotherapy were introduced in 2004, at the same time that the Cancer Registry underwent a transition to electronic reporting. Therefore, we chose only to include patients diagnosed from 2004 onward. Additionally, only the first hepato-pancreatico-biliary cancer was considered for each patient to avoid misclassifying regional spread as incident cancers.

Study design

All patients were followed from the date of their resection, and in case of multiple resections, the first was chosen. Patients were followed until the first of the following events occurred: death (the outcome of interest), emigration (an administrative censoring event), or end-of-study on 31 December 2023 (an administrative censoring event).

In Denmark, resection is only offered with curative intent, not for tumor debulking. Therefore, any resection performed was assumed to be with curative intent, and patients were assumed to have localized resectable disease at the time of resection. While registry data lack TNM staging, the type of surgery (liver resection, biliary tract resection, and combined resection) serves as a proxy for locoregional disease, consistent with Danish multidisciplinary team (MDT) standards since 2008[10]. In general, the resection type offered to patients with cholangiocarcinoma is based on tumor location within the biliary tree: Patients with intrahepatic cholangiocarcinoma are treated with liver resection; patients with perihilar cholangiocarcinoma are treated with a combination of liver resection and biliary tract resection; and patients with distal cholangiocarcinoma are treated with biliary tract resection (including Whipple’s procedure)[11].

Calculations

Comorbidities

To assess comorbidities, we computed the Charlson Comorbidity Index (CCI) score based on diagnoses recorded during the 5 years preceding resection. The score was computed based on the algorithm of Quan et al[12] who adapted the original score to fit the ICD-10 system[13]. Cancer and metastases were not included in the score. Patients’ scores were summarized as 0, 1, and ≥2. Chi-squared test of homogeneity was used to compare CCI scores across periods.

Resections

We calculated the proportion of patients who underwent liver resection, biliary tract resection, or both (on the same day), within 1 year of their diagnosis (the latest possible date of diagnosis was end December 2022, the latest date of resection was end December 2023). The proportion was calculated as the percentage of patients who were resected among all patients diagnosed with cholangiocarcinoma in each time period (2004–2012, 2013–2017, and 2018–2022). Chi-squared tests of homogeneity were used to test the hypothesis that the proportion undergoing resection was constant, one test per type of resection.

We chose to classify cholangiocarcinoma location based on the resection done, not on the diagnosis code recorded. This decision was based on our validation study, which showed that a patient’s first diagnosis code of cholangiocarcinoma frequently misclassified the anatomic location[4]. Importantly, the same validation study showed that 86% of patients with a diagnosis code for cholangiocarcinoma did in fact have a cholangiocarcinoma, and the remaining 14% had suspicion of cholangiocarcinoma, or they had another hepato-pancreatico-biliary cancer.

Chemotherapy

We computed the proportions of patients who received downstaging or adjuvant chemotherapy. We included all relevant chemotherapy regimens utilizing one of the four main drugs used in Denmark during the study period, alone or in combination: cisplatin, oxaliplatin, gemcitabine, or capecitabine. Downstaging was defined as treatment with any of the regimes between cholangiocarcinoma diagnosis and resection, while adjuvant treatment was defined as treatment with any of the regimes commenced from the day of resection and until 16 weeks (112 days) after resection^[11,14]. We did not have data on the intention of the chemotherapy treatment. Chi-squared test of homogeneity was used to compare proportions receiving chemotherapy across periods. Procedure codes and detailed definitions are listed in Supplemental Digital Content material 1, available at: http://links.lww.com/JS9/F18.

Mortality risk

Patients were followed from resection to death, and they were categorized by type of resection and calendar year of diagnosis (2004–2012, 2013–2017, and 2018–2022; aggregated due to small numbers of resectable patients). No patients died on the day of resection.

We used a Cox regression-based method to estimate age- and sex-adjusted 90-day, 1-year, and 3-year mortality risks from the time of resection. We used the overall age- and sex-distribution to adjust the mortality risk estimates for each year. Analyses were carried out in R as described in section 5.2 of the paper by Therneau et al using the adjcurv R-package^[4,15]. We computed the mortality risk difference of the age- and sex-adjusted risk estimates in 2018–2022 vs. 2004–2012 and used Fisher’s exact test to examine whether the difference was statistically significant.

Sensitivity analyses

We used sensitivity analysis to explore uncertainties in the definition of adjuvant treatment. First, patients need to have recovered satisfactorily after the resection before adjuvant chemotherapy is commenced, so we changed the denominator to patients alive 30 days after resection. Second, chemotherapy starting a long time after resection may in fact be palliative (not adjuvant) chemotherapy, so we restricted the numerator to patients who commenced adjuvant treatment within 12 weeks (84 days) after resection. Third, to approximate the influence of palliative treatments on our estimates of adjuvant chemotherapy, we computed the proportion of patients who received capecitabine in 2020–2022, after the change in guidelines following the BILCAP trial[14].

We also used three sensitivity analyses to explore time trends within subgroups of the cohort.

First, to examine if our estimates were influenced by changes in comorbidity burden across year of diagnosis, we repeated the analyses of mortality risk and used Cox regression to estimate confounder-adjusted time trends adjusted for (continuous) and within patient strata defined by the CCI score^[16,17] (CCI = 0 or CCI ≥ 1). Additionally, we examined the statistical interaction between year of diagnosis and CCI score by comparing Cox regression models with vs. without including the interaction term[18] (using the likelihood ratio test).

Second, surgical type was considered a disease stage proxy of locoregional disease, consistent with Danish MDT standards since 2008[10]. To examine robustness of our estimates, we repeated the mortality analysis only considering patients with TNM stage data recorded at the time of diagnosis, and with the stage recorded as non-metastatic.

Third, the biliary tract resections group covered resections of different extents. Therefore, we examined mortality risk in patients who specifically underwent Whipple’s procedure, which was a subset of the patients who underwent biliary tract resections, as this subgroup may be of particular interest.

Results

We included 611 patients with cholangiocarcinoma who underwent resection between 2004 and 2022, and they constituted 13% of the 4572 total patients diagnosed with cholangiocarcinoma. Of the 611 resected patients, 256 (42%) underwent liver resection, 249 (41%) underwent biliary tract resection (66% of biliary tract resections were part of a Whipple’s procedure), and 106 (17%) underwent combined liver and biliary tract resection. Women constituted 43% of patients throughout the study period. The median age of the patients who underwent resection increased from the first period (2004–2012) to the last period (2018–2022): for liver resection from 64 to 69 years, for biliary tract resection from 66 to 70 years, and for combined liver and biliary tract resection, age was stable around 65 years (Table 1).

Table 1.

Characteristics of patients with cholangiocarcinoma, dependent on type of resection and year of diagnosis, Denmark 2004–2022

	Number of patients	Females, number (%)	Age, median (IQR)
	Any resection
All	611	265 (43%)	67 (58–73)
2004–2012	155	67 (43%)	65 (53–71)
2013–2017	208	90 (43%)	67 (59–72)
2018–2022	248	108 (44%)	69 (60–74)
	Liver resection
All	256	124 (48%)	67 (59–73)
2004–2012	67	33 (49%)	64 (53–71)
2013–2017	68	32 (47%)	68 (62–74)
2018–2022	121	59 (49%)	69 (60–74)
	Biliary tract resection
All	249	94 (38%)	67 (59–74)
2004–2012	63	22 (35%)	66 (54–73)
2013–2017	94	36 (38%)	66 (59–71)
2018–2022	92	36 (39%)	70 (60–75)
	Combined resectiona
All	106	47 (44%)	65 (58–71)
2004–2012	25	12 (48%)	65 (55–69)
2013–2017	46	22 (48%)	66 (58–71)
2018–2022	35	13 (37%)	65 (59–71)

IQR, interquartile range.

^aCombined liver and biliary tract resection (both on the same day).

The proportion of patients who had comorbidities prior to their resection increased during the study period from 17% [95% confidence interval (CI) 12–24] in 2004–2012 to 28% (95% CI 23–34) in 2018–2022, P-value = 0.06. The proportion who had a CCI score of 1 was 13% (95% CI 8–19) in 2004–2012, while the proportion had increased to 19% (95% CI 15–25) in 2018–2022. The proportion of patients with a CCI score of 2 or more increased as well, from 5% (95% CI 2–9) in 2004–2012 to 8% (95% CI 6–13) in 2018–2022 (Supplemental Digital Content, Table 1, available at: http://links.lww.com/JS9/F18).

Resections

The proportion of patients with cholangiocarcinoma who underwent resection increased from 9% (95% CI 8–11) of patients diagnosed in the first period (2004–2012) to 16% (95% CI 14–17) of patients diagnosed in the last period (2018–2022), P-value < 0.01. The proportion who underwent liver resection increased from 4% (95% CI 3–5) in 2004–2012 to 8% (95% CI 6–9) in 2018–2022, P-value < 0.01, the proportion who underwent biliary tract resection increased slightly from 4% (95% CI 3–5) in 2004–2012 to 6% (95% CI 5–7) in 2018–2022, while the proportion who underwent a combined liver and biliary tract resection remained stable at 2–3%. The variation was statistically significant (P-value < 0.01) (Table 2).

Table 2.

The proportion of patients resected within 1 year of diagnosis of any cholangiocarcinoma, with 95% confidence interval, dependent on year of diagnosis, Denmark 2004–2022

Years	Cholangiocarcinoma
	Patients diagnosed	Any resection	Liver resection	Biliary tract resection	Combined resectiona
All	4572	13% (12–14)	6% (5–6)	5% (5–6)	2% (2–3)
2004–2012	1664	9% (8–11)	4% (3–5)	4% (3–5)	2% (1–2)
2013–2017	1319	16% (14–18)	5% (4–6)	7% (6–9)	3% (3–5)
2018–2022	1589	16% (14–17)	8% (6–9)	6% (5–7)	2% (2–3)
Chi-squaredb		<0.01	<0.01	<0.01	<0.01

^aCombined liver and biliary tract resection (both on the same day);

^bChi-squared test for homogeneity was used to compare proportions across all time periods.

Chemotherapy

Among the 611 patients who underwent resection, 9% (95% CI 7–11) had received downstaging chemotherapy and 40% (95% CI 34–46) received adjuvant chemotherapy. In general, the use of chemotherapy increased markedly from 2004–2012 to 2013–2017 and remained stable through the 2018–2022 period.

Downstaging chemotherapy was used increasingly before liver resection, from <7% in 2004–2012 to 12% (95% CI 8–20) in 2018–2022, P-value = 0.17. Among patients who underwent biliary tract resection, very few had received downstaging chemotherapy across all periods, 4% (95% CI 2–7). Among patients who underwent combined liver and biliary tract resection, the proportion who had received downstaging chemotherapy increased from none in 2004–2012 to 26% (95% CI 14–43) in 2018–2022 (Table 3).

Table 3.

The proportion of patients with cholangiocarcinoma who received chemotherapy, dependent on type of resection, timing of chemotherapy, and year of diagnosis, Denmark 2004–2022

Years	Downstaging	Adjuvant
	Liver resection
All	12% (8–16)	40% (34–46)
2004–2012	< 7%c	27% (18–39)
2013–2017	16–22%c	38% (27–50)
2018–2022	12% (8–20)	49% (40–58)
Chi-squaredb	P-value = 0.17	P-value = 0.01
	Biliary tract resection
All	4% (2–7)	48% (42–54)
2004–2012	0%	17% (10–29)
2013–2017	5–10%c	57% (47–67)
2018–2022	< 5%c	59% (48–68)
Chi-squaredb	P-value = 0.03	P-value < 0.01
	Combined resectiona
All	13% (8–21)	42% (33–52)
2004–2012	0%	< 20%c
2013–2017	11% (5–24)	52–61%c
2018–2022	26% (14–43)	49% (33–65)
Chi-squaredb	P-value = 0.01	P-value < 0.01

Note that patients may have received chemotherapy both before and after their resection and were then counted in both groups.

^aCombined liver and biliary tract resection (both on the same day);

^bChi-squared test for homogeneity was used to compare proportions across all time periods.

^cTo preserve patient anonymity, cells that either directly or indirectly reference less than five individuals are censored.

Adjuvant chemotherapy was used increasingly across all resection types: among patients who underwent liver resection, from 27% (95% CI 18–39) in 2004–2012 to 49% (95% CI 40–58) in 2018–2022, P-value = 0.01; among patients who underwent biliary tract resection, from 17% (95% CI 10–29) in 2004–2012 to 59% (95% CI 48–68) in 2018–2022, P-value <0.01; and among patients who underwent combined liver and biliary tract resection, from less than 20% in 2004–2012 to 49% (95% CI 33–65) in 2018–2022, P-value < 0.01 (Table 3).

Mortality risk

Among all patients who underwent resection, the mortality hazard ratio for a 5-year increase in age was 1.08 (95% CI 1.03–1.13), and for female compared with male sex it was 0.79 (95% CI 0.65–0.96). Because age was not a strong predictor of mortality, and because the sex distribution was constant during the study period, the crude and adjusted mortality risk estimates were virtually identical (Supplemental Digital Content, Table 2, available at: http://links.lww.com/JS9/F18).

In patients who underwent liver resection, 90-day, 1-year, and 3-year age- and sex-adjusted mortality were stable; 1-year adjusted mortality was 14% (95% CI 4–22) in 2004–2012 and 16% (95% CI 10–21) in 2018–2022, so the risk difference was 2 percentage points (P-value = 0.37). Likewise, in patients who underwent biliary tract resection, 90-day, 1-year, and 3-year adjusted mortality were all stable, e.g., the 1-year adjusted mortality was 24% (95% CI 12–33) in 2004–2012 and 21% (95% CI 12–29) in 2018–2022.

In the smaller group of patients who underwent a combined liver and biliary tract resection, by contrast, the adjusted mortality decreased. For example, the 1-year adjusted mortality decreased from 32% (95% CI 13–51) in 2004–2012 to 14% (95% CI 2–24) in 2018–2022. However, the decrease happened between 2004–2012 and 2013–2017, with a stable mortality risk from 2013–2017 to 2018–2022 (Fig. 1 and Table 4).

Figure 1.

Adjusted mortality, age and sex adjusted, dependent on type of resection, by aggregated calendar years of diagnosis, Denmark 2004–2022. Numbers along the top axis denote the number resected in each period.

Table 4.

Mortality risk, age- and sex-adjusted and with 95% confidence interval, stratified on resection type, and year of diagnosis, aggregated calendar years, Denmark 2004–2022

Years	Liver resection	Biliary tract resection	Combined resectiona
	90-day mortality risk (95% CI)
2004–2012	5% (0–9)	6% (0–12)	17% (1–31)
2013–2017	4% (1–8)	6% (1–12)	6% (0–14)
2018–2022	4% (1–8)	3% (0–6)	5% (0–11)
Differenceb	−1% point (−6 to 5), P-value = 0.42	−3% points (−10 to 4), P-value = 0.18	−11% points (−27 to 5), P-value = 0.09
	1-year mortality risk (95% CI)
2004–2012	14% (4–22)	24% (12–33)	32% (13–51)
2013–2017	20% (11–29)	19% (11–28)	17% (4–29)
2018–2022	16% (10–21)	21% (12–29)	14% (2–24)
Differenceb	2% points (−9 to 12), P-value = 0.37	−3% points (−16 to 10), P-value = 0.34	−18% points (−40 to 3), P-value = 0.05
	3-year mortality risk (95% CI)
2004–2012	53% (41–64)	56% (44–67)	74% (57–90)
2013–2017	42% (31–54)	66% (58–75)	43% (29–58)
2018–2022	44% (34–53)	56% (44–66)	41% (22–58)
Differenceb	−9% points (−24 to 6), P-value = 0.13	0% points (−16 to 16), P-value = 0.49	−33% points (−57 to −9), P-value < 0.01

CI, confidence interval.

^aCombined liver and biliary tract resection (both on the same day).

^bDifference of 2018–2022 vs. 2004–2012, compared using Fisher’s exact test.

Number at risk: liver resection, 2004–2012 = 67, 2013–2017 = 68, 2018–2022 = 121; biliary tract resection, 2004–2012 = 63, 2013–2017 = 94, 2018–2022 = 92; and combined resection, 2004–2012 = 25, 2013–2017 = 46, 2018–2022 = 35.

Sensitivity analyses

We repeated the analysis of adjuvant chemotherapy, including only patients alive 30 days after resection. This did not have any notable effect on the proportion who received adjuvant therapy (Supplemental Digital Content, Table 3, available at: http://links.lww.com/JS9/F18). Next, we identified patients who commenced adjuvant therapy within 12 weeks (instead of 16 weeks) and repeated the analysis. This also did not affect the proportion who received adjuvant chemotherapy (Supplemental Digital Content, Table 4, available at: http://links.lww.com/JS9/F18). Finally, we assessed the proportion of patients who received capecitabine as adjuvant treatment between 2020 and 2022 and found proportions similar to the overall estimates of 2018–2022. In 2020–2022, capecitabine was used as adjuvant treatment in 51% (95% CI 40–62) after liver resection, 46% (95% CI 34–60) after biliary tract resection, and 48% (95% CI 29–68) after a combined resection.

Considering the CCI score, we observed lower 1-year mortality with increasing diagnosis year, mainly in the group without comorbidities (Supplemental Digital Content, Table 5, available at: http://links.lww.com/JS9/F18). Accordingly, hazard ratios for calendar year were below 1, indicating a declining mortality over the study period. Interestingly, this declining mortality trend seemed to be stronger among patients with more severe comorbidity, although not in patients undergoing a combined resection. However, all time trends were imprecisely estimated due to the small number of patients, and P-values for the interaction between year of diagnosis and the CCI score were not statistically significant (Supplemental Digital Content, Table 6, available at: http://links.lww.com/JS9/F18).

When only considering patients with TNM data showing non-metastatic disease, results were generally similar to the main analysis, except that 1-year mortality risk in patients who underwent liver resection showed a stronger decreasing trend (Supplemental Digital Content, Table 7, available at: http://links.lww.com/JS9/F18).

Comparing 1-year mortality between patients who underwent a biliary tract resection with patient in which the biliary tract resection was a Whipple’s procedure (a subset of biliary tract resected patients), the estimates were similar [1-year mortality in 2018–2022: biliary tract resection, 21% (95% CI 12–29) and Whipple’s procedure 24% (95% CI 13–34)], and no trend was observed over time in either group (Supplemental Digital Content, Table 8, available at: http://links.lww.com/JS9/F18).

Discussion

Among the 4572 patients identified with cholangiocarcinoma between 2004 and 2022, 611 patients (13%) underwent resection. Overall, the proportion of patients who underwent resection increased from 9% of patients diagnosed in 2004–2012 to 16% of patients diagnosed in 2018–2022 – also the median age of resected patients and their comorbidity burden increased over time. We saw an increase in the proportion of patients with cholangiocarcinoma who were treated with liver resection alone (presumably for intrahepatic cholangiocarcinoma) or biliary tract resection alone (presumably for distal cholangiocarcinoma). Among these patients, more had received downstaging therapy, and more patients received adjuvant chemotherapy, but post-resection mortality did not change. We saw no increase in the proportion of patients with cholangiocarcinoma who were treated with combined liver and biliary tract resection (presumably for perihilar cholangiocarcinoma), but more of those who were treated had received downstaging chemotherapy, and more of them received adjuvant chemotherapy. In this group of patients treated with combined resection, mortality decreased from 2004–2012 to 2013–2017, but not thereafter.

The cancer care pathways in Denmark were introduced in 2008, providing regulated time limits for diagnosis and initiation of cancer treatment[10]. The Danish health authorities register all radiological examinations, and during our study period their usage has increased in number[19], with increasingly sensitive diagnostic tools being used (MRI, CT, MRCP, etc.). This increased use and sensitivity of diagnostic tests was one of the two major changes during the study period, the other being the developments in chemotherapy, which may have resulted in changes to patients’ characteristics and comorbidity burden, and at the same time may help explain our findings.

First, regarding the sensitivity and use of diagnostic tests, the cancer care pathways may explain the increase in the proportion of patients who had localized resectable cholangiocarcinoma at diagnosis, which was observed in our previous study[4]. The same study showed a sharp increase in the incidence of intrahepatic cholangiocarcinoma in Denmark[4] – in fact, the 2-fold increase in the number of patients receiving hepatic resection is dwarfed by the 5-fold increase in the number of patients diagnosed with intrahepatic cholangiocarcinoma. Moreover, our previous study showed an increasing proportion of patients whose intrahepatic cholangiocarcinoma was localized at the time of diagnosis, and also an increasing chance of undergoing resection if you did have such an early diagnosis[4]. The best way to reconcile these findings with what we show here is that (1) the cancer care pathways result in more diagnoses of early-stage intrahepatic cholangiocarcinoma, yes, but they also result in more diagnoses of late-stage intrahepatic cholangiocarcinoma; (2) many patients with early-stage intrahepatic cholangiocarcinoma are not fit for surgery; and (3) as detailed in the next paragraph, downstaging chemotherapy has made previously unresectable cancers resectable. Undoubtedly, the characteristics of patients who undergo surgery for cholangiocarcinoma have changed considerably during our study period, as was evident from the increase in median age and comorbidity burden.

Second, regarding the introduction of chemotherapy, in 2010 following the ABC-02 trial[20], the combination of cisplatin and gemcitabine was introduced as first-line treatment in patients with non-resectable or metastatic cholangiocarcinoma in Denmark[11]. The increased proportion of resected patients who have received downstaging chemotherapy likely stems from downstaging chemotherapy having enabled previously unresectable patients to undergo resection after treatment with cisplatin and gemcitabine[11]. These borderline resectable patients had more advanced disease at the time of diagnosis, compared with clearly resectable patients, and therefore also a worse prognosis per se. This might help explain why no decrease in mortality risk was observed. Furthermore, the introduction of adjuvant chemotherapy may have contributed to a decrease in mortality following resection. Following the publication of the BILCAP trial[14] in 2019, capecitabine was introduced in Denmark as standard adjuvant therapy, recommended to all patients with cholangiocarcinoma and commenced within 16 weeks following resection^[11,14]. We found that 52% (95% CI 46–59) received adjuvant therapy in 2018–2022, and the proportion was similar if only considering 2020 onward at 59% (95% CI 51–66), or if only considering patients treated with capecitabine in 2020–2022 at 49% (95% CI 41–57). Adjuvant chemotherapy is commenced as soon as possible after adequate wound healing and complete pathology assessment, but we find that 59% receipt of adjuvant treatment is compatible with our clinical experience: Many patients do not receive adjuvant chemotherapy due to poor performance status after resection, surgical complications, and/or infection. In summary, we believe that the stable mortality risk despite chemotherapy adoption reflects a combination of multiple factors: (1) expanded resection criteria including borderline-resectable cases, (2) reduced chemotherapy tolerance, or preference, in older patients (median age increased by 4 years during our study period)[14], and (3) counterbalancing effects of postoperative complications, which is an important aspect to examine in future studies.

While more patients are resected, mortality risks have remained stable. We saw a stable mortality risk in all patients undergoing resection, independent of resection type, in the last 10 years (2013–2017 to 2018–2022). The stable mortality risk may be caused by opposing factors influencing the estimate: (1) Earlier diagnosis may have allowed previously unresectable patients to undergo resection and may additionally have decreased the mortality risk after resection; (2) downstaging chemotherapy may have broadened the indication for resection, allowing resection of patients previously deemed irresectable; (3) improvements in surgical procedures and postoperative management, including an increased use of adjuvant therapy, may have broadened the indication for surgery and affected outcomes after surgery. The increase in median age and comorbidity burden of resected patients indicates that patients were more frail in later years than patients resected previously; and (4) finally, the mortality risk estimates had wide CIs, and the uncertainty suggests that chance may also have influenced estimates, while earlier identification of patients may have caused lead time bias. Unfortunately, we are not able to disentangle the effects of these multiple factors.

When examining the impact of comorbidity, we might observe a trend toward greater reduction in mortality over time for patients with a greater burden of comorbidity, although there were not enough patients within strata to make a judgment. If reduction in mortality was greater with greater burden of comorbidity, such a finding may be explained by improvements in surgical techniques and post-operative care having a larger impact on patients with more comorbidities.

The validity of our data is corroborated by an independent data source. In Denmark, in addition to the National Patient Registry and the Cancer Registry, we have the Danish Liver-Biliary Tract Cancer Database (DLGCD) administered by the Danish Liver Cancer Group[21]. This database contains manually entered data on all patients with hepatocellular carcinoma, intrahepatic or perihilar cholangiocarcinoma, or gallbladder cancer who have been referred to an MDT conference[22]. The findings of our study correspond well with estimates reported in the latest annual report by this group. First, we found similar resection proportions. In the annual report, they report that among 1111 patients diagnosed with biliary tract or gallbladder cancer in 2021–2023, 217 underwent resection or ablation with resections constituting 82–83% of the 217 treatments, i.e., (0.825 * 217)/1111 = 16% of the patients seen on MDT were resected[22] – we found the same proportion. Second, we found comparable 1-year mortality after resection. In the report, they estimate that among 51 patients with biliary tract or gallbladder cancer who received first-time local treatment in 2022, 1-year mortality was 16% (95% CI 7–29), and they did not observe a trend in mortality risk between 2015 and 2022[22]. This was similar to our finding of a stable 1-year mortality between 14% and 21% in 2012–2022. Third, we found reasonably similar proportions treated with downstaging chemotherapy. They reported that among all 1111 patients, 3% received downstaging chemotherapy[22]. Among patients who were resected, we found that 9% had received downstaging chemotherapy, which was, as would be expected, higher than the reported 3% in DLGCD among all diagnosed patients. When comparing the annual report from the DLGCD with our study, it is worth noting that the aim of this study was to describe time trends in mortality from resection, whereas the DLGCD was established as a quality control measure. Further, unlike the DLGCD, we included distal cholangiocarcinomas, whereas we did not report on gallbladder cancer. We believe that the high level of agreement between our findings and those reported by the Danish Liver Cancer Group in their annual report indicates that our data have high validity.

We aimed to describe changes in the treatment of cholangiocarcinoma and mortality risk following resection. We could not provide a clear explanation for the changes we described, and that was a limitation of our study. To improve the outcome for patients with cholangiocarcinoma, future research should focus on more granular data collection to determine the real-world impact of treatment advancements. The modern treatment of cholangiocarcinoma is precision-guided, and the information in most registries today is not sufficiently granular to capture important variation across patients and tumors. Moreover, clinically relevant data should be collected at multiple times during the disease course, e.g., at the time of diagnosis, at the time of resection, and at the initiation of systemic treatment.

The major limitation of this study was that we lacked relevant clinical data from the time of resection, e.g., data on stage, performance status, frailty, biochemistry, resection margins, and complications, and so we could not assess whether tumors and patients were comparable across the study period. Further, unmeasured stage migration may have confounded temporal comparisons[23]. Even so, our study provides an unparalleled nationwide overview of clinical practice and outcomes for patients with cholangiocarcinoma over two decades that have seen major changes in clinical guidelines, diagnostic tools, and access to care.

In conclusion, between 2004 and 2022, the proportion of patients with cholangiocarcinoma who underwent liver or biliary tract resection increased, and more patients received downstaging chemotherapy and adjuvant chemotherapy. Nonetheless, age- and sex-adjusted mortality risks remained constant, except for a decreasing mortality for patients treated with combined resection 2013–2017 vs. 2004–2012. The explanations for the time trends we report are manifold and have likely resulted from changing patient characteristics.

Ethical approval

This study did not require approval from an ethics committee according to Danish law.

Consent

This study did not require written consent from patients according to Danish law.

Sources of funding

P.J.’s work is supported by a grant from the Novo Nordisk Foundation (grant reference number: NNF19OC0054612). M.D.J.’s work is supported by a grant from Aarhus University. The funding sources had no involvement in the study.

Author contributions

All authors contributed to the study design and method selection. M.D.J. conducted the analyses of the data supervised by P.J. M.D.J. wrote the first draft of the paper, and all authors critically revised the manuscript and approved the final version.

Conflicts of interest disclosure

The authors declared that no conflict of interest.

Research registration unique identifying number (UIN)

This study did not recruit patients, but used routinely collected pseudonymized data.

Guarantor

Morten Daniel Jensen.

Provenance and peer review

Not commissioned, externally peer-reviewed.

Data availability statement

Electronic health records are considered “sensitive” data in Denmark and cannot be shared. However, data can be accessed through an application to the Danish Health Data Authority.

Assistance with the study

None.

Presentation

None.