Liver Blood Tests in the Management of Suspected Choledocholithiasis

Abstract

Objective

The likelihood of common bile duct (CBD) stones considers liver blood tests (LBTs) if they are markedly altered only. The aim of our study was to find a reliable tool based on LBTs to predict the presence of CBD stones.

Methods

We retrospectively considered all patients who underwent magnetic resonance cholangiopancreatography (MRCP) because of suspected CBD stones from January 2014 to June 2019. Demographic, clinical data, and LBT values were collected and analyzed.

Results

We selected 191 patients, 64 (33.5%) with positive MRCP and 127 (66.5%) with negative MRCP. The analysis showed that our compound LBT-based score had 83.6%, 90.7%, and 90.6% sensitivity, specificity, and negative predictive values, respectively, in determining MRCP results.

Conclusion

We designed a weighted score with high diagnostic power in determining MRCP results that could help in differentiating between candidates for primary cholecystectomy and patients who benefit from preoperative MRCP.

cholestasis, common bile duct gallstones, magnetic resonance cholangiopancreatography, gallbladder, predictive value of tests, cholecystectomy

Gallstones are a common disease worldwide, affecting up to 21% of the general population.¹ Approximately 35% of patients with gallstones develop symptoms or complications.² Medical conditions such as biliary colic, jaundice, pancreatitis, cholecystitis, and cholangitis are common indications for hospital admission.³ Up to 15% of patients with symptomatic gallstone disease have common bile duct (CBD) stones, which may be asymptomatic in 5% to 12% of patients or lead to more severe conditions.⁴

Patients admitted to the hospital with biliary colic are usually evaluated with clinical history, physical examination, blood tests, and abdominal ultrasound,⁵ not only to confirm the diagnosis of gallstones but also to evaluate the risk of CBD stones, because missing their occurrence can result in complications such as cholangitis, acute biliary pancreatitis, and symptomatic choledocholithiasis.⁵ The guidelines of the European Association for the Study of the Liver support the routine use of serum liver biochemical tests and abdominal ultrasound as the initial evaluation to stratify the risk of CBD stones. American, European, British, and German societies’ guidelines suggest predictors for CBD stones and propose management algorithms.^6-9 Most societies advocate preoperative endoscopic retrograde cholangiopancreatography (ERCP) in patients with a high likelihood of CBD stones, whereas magnetic resonance cholangiopancreatography (MRCP), endoscopic ultrasonography (EUS), or intraoperative cholangiography are indicated in case of intermediate probability. Finally, patients with a low probability of CBD stones should undergo a cholecystectomy with or without an intraoperative cholangiography.

Therefore, clinical suspicion, abdominal ultrasound, and liver blood tests (LBTs) play the main role in predicting the presence of CBD stones. However, none of them has an adequate sensitivity or specificity when used alone, and the recent literature is still controversial.^5,10,11 The aim of our study was to develop a mathematical equation based on LBT values and to use it as a score to predict the presence of CBD stones.

Materials and Methods

Data Collection

Written consents were administered to all patients, and the local ethics committee approved the study (Comitato Etico Cantonale Ticino number 2019–02060 CE 3535) in accordance with the principles set forth in the Helsinki Declaration.

At our institution, we retrospectively evaluated all surgical patients affected from biliary colic who underwent MRCP from January 2014 to June 2019. Patients with a past history of ERCP and/or biliary stent and/or known hepatobiliary pancreatic neoplastic lesions or known primary sclerosing cholangitis were excluded. The dataset included age, sex, symptoms, and LBTs. Specifically, we collected total (TotBil, mg/dL), direct bilirubin (DirBil, mg/dL), aspartate aminotransferase (AST, U/L), alanine aminotransferase (ALT, U/L), gamma-glutamyltransferase (GGT, U/L), and alkaline phosphatase (AP, U/L). The LBT values were recorded on admission day (@time0) and 2 days later (@time1). Afterward, we recorded the variation in the values of such parameters over time as delta scores (Δ): ΔTotBil, ΔDirBil, ΔAST, ΔALT, ΔGGT, and ΔAP.

Statistical Analysis

For statistical software, we used MedCalc Statistical Software version 19.7 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org; 2021). An analysis of variance test was used to evaluate demographic, clinical outcome, and LBTs differences between groups. For all LBTs at different time points, a receiver operating characteristic (ROC) curve,¹² sensitivity, specificity, positive predictive value, and negative predictive value were calculated. The Youden’s index was used to establish the best threshold on the ROC curve.¹³ A logistic regression analysis was used to find the best area under the ROC curve (AUC) value, single coefficients were therefore collected, and an LBT-weighted score was calculated (LBT_score). The K-folds cross-validation method (K = 5) was used to create multiple validation subsets of our data sample and to assess our prediction model reliability.¹⁴ The sample size calculation set an amount of 149 patients to achieve 80% power with a type I error of 0.05. The threshold of statistical significance was P < .05.

Data Availability

The data associated with this article are not publicly available but are available from the corresponding author on reasonable request.

Results

Over the study period, out of 224 identified patients, 33 were excluded (26 patients had known hepatobiliary pancreatic neoplastic lesions and 7 patients suffered from primary sclerosing cholangitis; Figure 1). The demographic and clinical features of the 191 patients who matched the inclusion and exclusion criteria are shown in Table 1.

Table 1.

Demographic and Clinical Characteristics of Patients Included in the Study

	Positive MRCP n = 64	Negative MRCP n = 127	P
Age (y, SD)	67.2 (16.9)	60.1 (18.1)	.009
Male (n, %)	34 (53)	53 (42)	.368
Abdominal pain (n, %)	47 (73)	105 (83)	.611
Acute cholecystitis (n, %)	5 (8)	16 (13)	.369
Acute pancreatitis (n, %)	10 (16)	30 (24)	.295
Jaundice without symptoms (n, %)	15 (23)	1 (1)	<.001
Altered LBTs without symptoms (n, %)	6 (9)	20 (16)	.287
Increased LBTs (%)
TotBil (n, %)	48 (75)	74 (50)	.294
DirBil (n, %)	49 (76)	67 (53)	.125
AST (n, %)	48 (75)	82 (65)	.592
ALT (n, %)	48 (75)	92 (72)	.883
GGT (n, %)	56 (87)	99 (78)	.611
AP (n, %)	57 (89)	102 (80)	.647

ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase; DirBil, direct bilirubin; GGT, gamma-glutamyltransferase; LBTs, liver blood tests; MRCP, magnetic resonance cholangiopancreatography; SD, standard deviation; TotBil, total bilirubin.

Values are expressed as number of patients and percentage of the total.

Figure 1

Flow chart of patients’ selection and study methodology. MRCP, magnetic resonance cholangiopancreatography.

The MRCP was negative in 127 patients (66.5%). Among the 64 patients (33.5%) with a positive MRCP finding, 50 of them (26.2%) had 1 or more stones, whereas 14 patients (7.3%) had no stones but lesions of another nature that required further investigations.

Results on ROC analyses are reported in Table 2. Among all LBTs, DirBil at MRCP was found to have the best diagnostic power in predicting MRCP result (73.7%, 88.8%, 76.4%, and 86.3% sensitivity, specificity, positive predictive value, and negative predictive values, respectively). Coefficients of the logistic regression analysis were used as follows:

Table 2.

ROC Analyses Showing the Diagnostic Power of Each LBT

LBT	Associated Threshold (Youden’s index)	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)	AUC	P
On hospital admission
TotBil (mg/dL)	>1.55	75.0	50.8	43.6	80.0	0.651	<.001
DirBil (mg/dL)	>0.91	68.7	62.7	41.4	79.8	0.675	<.001
AST (U/L)	>78	71.9	39.7	37.7	73.5	0.524	.592
ALT (U/L)	>40	85.9	23.8	36.4	76.9	0.507	.876
GGT (U/L)	>316	62.5	57.1	42.6	75.0	0.595	.030
ALP (U/L)	>183	59.4	69.1	49.4	77.0	0.647	.001
Day of MRCP
TotBil (mg/dL)	>2.34	71.9	84.5	69.5	86.0	0.813	<.001
DirBil (mg/dL)	>1.20	73.7	88.8	76.4	86.3	0.848	<.001
AST (U/L)	>80	70.2	66.4	50.6	81.9	0.702	<.001
ALT (U/L)	>361	31.6	88.8	58.1	72.5	0.613	.014
GGT (U/L)	>276	78.9	57.8	47.9	84.8	0.685	<.001
ALP (U/L)	>166	79.0	63.8	51.7	86.0	0.741	<.001
Δ values
TotBil (mg/dL)	<–0.15	52.6	84.5	62.5	78.4	0.695	<.001
DirBil (mg/dL)	<–0.01	59.6	83.6	64.2	80.8	0.720	<.001
AST (U/L)	<72	73.7	46.6	40.4	78.3	0.622	.007
ALT (U/L)	<125	86.0	36.2	39.8	84.0	0.614	.011
GGT (U/L)	<–1	42.1	75.9	46.2	72.7	0.579	.094
ALP (U/L)	<–50	33.3	90.5	63.3	73.4	0.581	.103
LBT weighted formula	>–0.98	83.6	90.7	82.7	90.6	0.890	<.001

ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase; AUC, area under the ROC curve; DirBil, direct bilirubin; GGT, gamma-glutamyltransferase; LBT, liver blood test; MRCP, magnetic resonance cholangiopancreatography; NPV, negative predictive value; PPV, positive predictive value; ROC, receiver operating characteristic; TotBil, total bilirubin.

Δ indicates the difference of all variables over the time period.

$${\displaystyle \begin{array}{l}{\displaystyle \begin{array}{lll}& {\displaystyle \mathrm{L}\mathrm{B}{\mathrm{T}}_{\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}}:-}& {\displaystyle 2.21347-0.00947\ast \mathrm{T}\mathrm{o}\mathrm{t}\mathrm{B}\mathrm{i}{\mathrm{l}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}0}-0.01218}\\ & & \ast \mathrm{D}\mathrm{i}\mathrm{r}\mathrm{B}\mathrm{i}{\mathrm{l}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}0}-0.00048\ast \mathrm{A}\mathrm{S}{\mathrm{T}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}0}-0.00041\\ & & \ast \mathrm{A}\mathrm{L}{\mathrm{T}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}0}-0.00160\ast \mathrm{G}\mathrm{G}{\mathrm{T}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}0}+0.00199\\ & & {\displaystyle \ast \mathrm{A}\mathrm{L}{\mathrm{P}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}0}-0.01971\ast \mathrm{T}\mathrm{o}\mathrm{t}\mathrm{B}\mathrm{i}{\mathrm{l}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}1}+0.11127}\\ & & \ast \mathrm{D}\mathrm{i}\mathrm{r}\mathrm{B}\mathrm{i}{\mathrm{l}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}1}{\displaystyle +0.00258\ast \mathrm{A}\mathrm{S}{\mathrm{T}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}}+0.00009}\\ & & \ast \mathrm{A}\mathrm{L}{\mathrm{T}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}1}+0.00184\ast \mathrm{G}\mathrm{G}{\mathrm{T}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}1}+0.00023\ast \mathrm{A}\mathrm{L}{\mathrm{P}}_{\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}1}\end{array}}\end{array}}$$

The AUC of the LBT_score was 0.890 (P < .001) and was the best value as compared to the single LBT (Figure 2). The score predicted 92.4% of patients correctly, all 14 patients with positive MRCP because of unclear lesions were classified as high risk, and further investigation was correctly suggested.

Figure 2

ROC curve analyses. ALT, alanine aminotransferase; AP, alkaline phosphatase; DirBil, direct bilirubin; GGT, gamma-glutamyltransferase; LBT, liver blood test; ROC, receiver operating characteristic; TotBil, total bilirubin.

The K-folds cross-validation method confirmed the high diagnostic power of our LBT_score prediction model. The sensitivity, specificity, positive predictive value, and negative predictive value were 83.6%, 90.7%, 82.7%, and 90.6%, respectively.

Discussion

This is the first report that combines LBT values and their trend over time (specifically over 48 hours) in a mathematical model. The LBT_score showed high diagnostic value in patients with suspected CBD stones, and 92.4% of MRCP findings could be correctly predicted.

When misdiagnosed, CBD stones can lead to severe complications.⁵ Therefore, several studies have tried to stratify patients with suspected CBD stones in different risk categories. American, European, British, and German societies’ guidelines, among others, suggest predictors for CBD stones along with management algorithms.^6-9 According to the guidelines of Williams et al,⁹ the likelihood of CBD stones should be established on history-taking, LBTs, and abdominal ultrasound findings. The American guidelines identify a high probability of CBD stones in patients with gallstones and visible CBD stones on ultrasound, symptoms and signs of cholangitis, or a combination of CBD dilatation and jaundice.⁸ Altered LBTs are helpful in stratifying the risk, although a quantitative threshold is not commonly used in clinical practice.^11-22 Patients admitted with biliary colic often present with abnormal LBTs, so that with inconclusive ultrasound findings, a preoperative diagnostic of the biliary tree is often performed to rule out CBD stones.

An MRCP is an excellent imaging tool to study the biliary tree; however, it is not the only available method. According to a recent Cochrane systematic review,²⁴ MRCP is highly sensitive (93%) and specific (96%) in diagnosing CBD stones. In addition, MRCP is widely available in developed countries, not invasive, and, unlike EUS, suitable in case of altered gastroduodenal anatomy. However, it cannot be used in patients with mechanical heart valves and pacemakers. Claustrophobia and obesity represent further limitations. An EUS shows high sensitivity (95%) and specificity (97%) in detecting CBD stones as well. Because of the high diagnostic value of such techniques, both can be considered in the preoperative diagnosis of CBD stones. We considered MRCPs as a reference, and the LBT_score should be used to select patients for such an investigation because preoperative EUS is not directly applicable.

Most societies recommend preoperative ERCP in patients with a high likelihood of CBD stones, whereas MRCP, EUS, or intraoperative cholangiography are indicated in intermediate probability.^6-9 Patients with a low probability of CBD stones should undergo a cholecystectomy with or without intraoperative cholangiography without further investigations.²⁴ In acute patients, carrying out unnecessary MRCP, ERCP, and EUS can delay cholecystectomy and increase the length of stay and the hospital costs.²⁵ In our series, we observed a large number of negative MRCPs, so we tried to find a more reliable parameter that could justify MRCP before cholecystectomy using, retrospectively, the LBT values of 191 patients and a mathematical model based on regression analysis. Thereby, in measuring the LBT values at time 0 (day of admission) and at time 1 (48 hours later) and combining these values according to the above-detailed formula, we found an interesting diagnostic tool that we named the LBT_score. Our score reliability was also confirmed by the K-folds cross-validation method.¹⁴ According to our results, in a patient admitted with biliary colic and observed for 48 hours, LBT_score > –0.98 would indicate a high probability of negative MRCP results. Once validated in a prospective study design, the LBT_score could be helpful in indicating whether preoperative MRCP, eventually followed by ERCP, or primary cholecystectomy is necessary.

This study has several limitations. A retrospective data collection was used to construct the predictive score. The relatively small number of patients does not allow strong conclusions but is nevertheless statistically significant because the sample size calculation set an amount of 149 patients to achieve 80% power with a type I error of 0.05.

In addition, our study does not predict the presence of choledocholithiasis but rather the result of MRCP. However, MRCP is an excellent imaging tool to study the biliary tree and an equivalence might be argued. In 2003, Topal et al²⁶ showed that MRCP has a positive predictive value of 100% and a negative predictive value of 98% in diagnosing CBD stones. According to a recent Cochrane systematic review,²⁴ MRCP is highly sensitive (93%) and specific (96%) in diagnosing CBD stones. In addition, MRCP is widely available in developed countries, is not invasive, and, unlike EUS, is suitable in case of altered gastroduodenal anatomy. In this context, our work can be considered as a pilot study to be integrated into a more comprehensive algorithm to predict the presence of CBD stones.

Conclusion

Patients with suspected CBD stones are usually managed according to stratification into low-, intermediate-, and high-risk groups. In such patients, planning and carrying out unnecessary MRCP and EUS can lead to a delay of surgical treatment, increasing the length of hospital stay and costs. We designed a score based on LBT values over a 48-hour time frame. Based on our retrospective data, the LBT_score showed a high diagnostic power in predicting MRCP results. Once the score is validated in a prospective study design, it may be helpful in selecting patients with suspected CBD stones who would definitely benefit from a preoperative MRCP.

Abbreviations

Abbreviations
 
• CBD

  common bile duct

 
• LBT

  liver blood test

 
• MRCP

  magnetic resonance cholangiopancreatography

 
• ERCP

  endoscopic retrograde cholangiopancreatography

 
• EUS

  endoscopic ultrasonography

 
• TotBil

  total bilirubin

 
• DirBil

  direct bilirubin

 
• AST

  aspartate aminotransferase

 
• ALT

  alanine aminotransferase

 
• GGT

  gamma-glutamyltransferase

 
• AP

  alkaline phosphatase

 
• ROC

  receiver operating characteristic

 
• AUC

  area under the ROC curve

Acknowledgments

The authors thank Professor Pietro Majno-Hurst for comments and suggestions. Written nonopposition consents were administered to patients, and the local ethics committee approved the study (Comitato Etico Cantonale Ticino number 2019–02060 CE 3535). Consent for publication was obtained from all patients.