Ultrasound Surveillance of Small, Incidentally Detected Gallbladder Polyps: Projected Benefits by Sex, Age, and Comorbidity Level

Abstract

Objective:

Incidentally detected gallbladder polyps are commonly encountered when performing upper abdominal ultrasound. Our purpose was to estimate the life expectancy (LE) benefit of ultrasound-based gallbladder surveillance in patients with small (6–7 to <10 mm), incidentally detected gallbladder polyps, accounting for patient sex, age, and comorbidity level.

Methods:

We developed a decision-analytic Markov model to evaluate hypothetical cohorts of women and men with small gallbladder polyps, with varying age (66–80 years) and comorbidity level (none, mild, moderate, severe). Drawing from current evidence, in the base case, we assumed no increased risk of gallbladder cancer in patients with small gallbladder polyps. To estimate maximal possible LE gains from surveillance, we assumed perfect cancer control consequent to 5 years of surveillance. We varied key assumptions including cancer risk and test performance characteristics in sensitivity analysis.

Results:

Projected LE gains from surveillance were <3 days across most cohorts and scenarios evaluated. For 66- and 80-year-olds with no comorbidities, LE gains were 1.46 and 1.45 days, respectively, for women, and 0.67 and 0.75 days for men. With 10 years of surveillance, LE gains increased to 2.94 days for 66-year-old women with no comorbidities (men: 1.35 days). If we assumed a 10% increase in gallbladder cancer risk among individuals with polyps, LE gains increased slightly to 1.60 days for 66-year-old women with no comorbidities (men: 0.74 days). Results were sensitive to test performance and surgical mortality.

Discussion:

Even under unrealistic, optimistic assumptions of cancer control, ultrasound surveillance of incidentally detected small gallbladder polyps provided limited benefit.

INTRODUCTION

Gallbladder polyps that are incidentally detected at ultrasound represent a challenging management dilemma for radiologists and referring physicians [1,2]. Incidentally detected gallbladder polyps are common, encountered at approximately 3% to 5% of upper abdominal ultrasound examinations [4,5]. Most are not “true” polyps—adenomatous or cancerous lesions—but rather “pseudopolyps,” representing foci of cholesterol deposition or artifacts of imaging [2]. With increasing polyp size, the likelihood of a true polyp and cancer is increased [6]. As such, size criteria are used to trigger surveillance [1,7–9,29]. Patients with polyps 1 to 1.5 cm or larger are typically referred for surgical management, given the risk of malignancy and the aggressive nature of gallbladder cancer [,7–9,29]. Most polyps <6 to 7 mm are presumed to be benign in etiology, and generally, no surveillance is pursued [7–9,29]. For polyps from 6–7 to <10 mm, surveillance is commonly recommended [7–9,29]. The 2020 Canadian Association of Radiologists recommendations [7] support surveillance for up to 5 years. More recent guidance (2022) from the Society of Radiologists in Ultrasound (SRU) [8] supports 1 year of surveillance or less for most small gallbladder polyps, depending on morphology. Updated (2022) European multisociety consensus recommendations [29] support an intermediate surveillance length of 2 years in most settings in which surveillance of small gallbladder polyps is pursued.

In real-world practice, the value of ultrasound surveillance for gallbladder polyps that are 6–7 to <10 mm is commonly questioned [10]. In a recent, large, retrospective cohort study, patients with small gallbladder polyps had no elevated risk of gallbladder cancer compared with patients without gallbladder polyps [6]. Even if the small, observed risk (<0.1%) of gallbladder cancer in the study population [6] could be averted by surveying the gallbladders of individuals with polyps within this size range, for many patients with incidentally detected gallbladder polyps, competing risks of comorbidities, advanced age, and surgical complications would further attenuate the small, hypothetical benefits of surveillance at the population level.

In this study, adhering to methods previously used to examine the benefits of surveillance of incidental findings in other settings [11,12], we projected the life expectancy (LE) benefits of small gallbladder polyp (6–7 to <10 mm) surveillance over 5 years. We sought to predict the maximum achievable LE benefits from surveillance by first assuming perfect gallbladder cancer detection and treatment. We also examined varied assumptions of imaging test performance for cancer detection, gallbladder cancer risk, and surveillance length, as well as age and comorbidity level at the time of polyp detection. Given the potential risks of overtreatment—as well as associated resource requirements—we sought to quantify the value of surveillance in this setting. Understanding the projected benefits of surveillance under different possible scenarios, as allowed by our study, is important when considering a substantial change in screening paradigm.

METHODS

Overview

We developed a decision-analytic Markov model to simulate patient cohorts with small, incidentally detected gallbladder polyps using TreeAge Pro software (Healthcare Version 2020, Williamstown, Massachusetts). We estimated the LE benefit of 5 years of surveillance with ultrasound imaging compared to no surveillance (Fig. 1). We used a 5-year surveillance period, varying the surveillance period in a sensitivity analysis to account for shorter surveillance, including recent SRU recommendations [8]. In practice, the duration of surveillance is recognized to be heterogeneous [10].

Fig. 1.

Model schematic. Cohorts of individuals defined by patient sex, age, and comorbidity level enter the model with a history of an incidental gallbladder polyp. They either undergo surveillance imaging or do not as detailed in this schematic.

Based on current evidence [6], we assumed that the risk of gallbladder cancer for a patient with a polyp was the same as that for a patient without a polyp. We also assumed that ultrasound identified polyps in the early stage of malignant transformation, thereby eliminating the risk of dying from gallbladder cancer during surveillance. In doing so, we estimated the theoretical maximum benefits of surveillance in the base-case analysis. In sensitivity analyses, we varied these critical assumptions to determine the effects of elevated cancer risk in individuals with polyps and of imperfect cancer detection at ultrasound surveillance on different management strategies.

The Markov model simulated patient cohorts from specified ages of incidental polyp detection (66–80 years) to the end of their life and accounted for patients’ risk of dying from gallbladder cancer or other causes. These cohorts of women and men were assigned a comorbidity level (none, mild, moderate, and severe). Medical conditions constituting each comorbidity level have been previously published [13]. We assumed gallbladder cancer outcomes were independent of comorbidity status, absent specific data to inform this relationship.

Management Strategies

Surveillance Strategy.

In the surveillance strategy, patients experienced 5 years of ultrasound imaging surveillance. In the base case, our model structure was such that all gallbladder cancers were detected and successfully treated (ie, patients were cured of cancer) during the surveillance period, sparing patients from gallbladder cancer-specific mortality during surveillance [14]. The interval of imaging, every 6 months versus every year, was therefore not relevant to our analysis. After the surveillance period, patients who did not have gallbladder cancer were at average risk of gallbladder cancer death. Additionally, patients were subject to the risk of death from causes other than gallbladder cancer [13,15,16] during and after the surveillance period.

No Surveillance Strategy.

In the no surveillance strategy, patients were subject to average risks of developing gallbladder cancer. We elicited relative survival rates for patients who developed gallbladder cancer from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program data [14]. All patients in the no surveillance strategy were subject to other-cause mortality risks in addition to the risk of gallbladder cancer-specific death [13,15,16].

Gallbladder Cancer Risk and Survival in the Model

Gallbladder Cancer Risk.

Patients with incidentally detected gallbladder polyps had the same risk of developing gallbladder cancer regardless of the management strategy. We used DevCan software applied to SEER data to estimate sex- and age-specific risks of developing gallbladder cancer (Fig. 2) [14,17,18]. SEER data included patients diagnosed with gallbladder cancer (C239-Gallbladder) from 1999 to 2005.

Fig. 2.

Age-conditional probability of developing gallbladder cancer by single age given cancer-free at the previous age from DevCan software applied to Surveillance, Epidemiology, and End Results program data for patients diagnosed between 1999 and 2005 [14,17,18]. Women have a higher risk of developing gallbladder cancer than men.

Gallbladder Cancer Survival.

Patients with gallbladder cancer were placed on an associated all-stage relative survival curve obtained from SEER data from 1992 to 2005 [14]. Our approach for obtaining all-stage relative survival estimates was the same as in prior studies [11,12].

Modeling Sex-, Age-, and Comorbidity-Specific Other-Cause Mortality Rates

Sex-, age-, and comorbidity-specific mortality rates from causes other than gallbladder cancer were adapted for use in our analysis as follows. Cancer-free comorbidity-specific mortality rates, developed using SEER-Medicare data from 1992 to 2005 [13,16], were previously adjusted to include risks of cancer death (all cancers) [15] using validated methods [11,12]. Using multiple cause of death data from the National Center for Health Statistics, we obtained sex- and age-specific rates of gallbladder cancer-specific mortality (underlying cause of death code: C23). Using these data, we adjusted our cancer-inclusive comorbidity-specific mortality rates to reflect sex-, age-, and comorbidity-specific mortality rates from causes other than gallbladder cancer, as was needed for our analysis.

Base-Case Analysis

Our base-case analysis measured the LE from the surveillance and no surveillance strategies stratified by patient sex, age, and comorbidity level, assuming perfect cancer detection by ultrasound (100% sensitivity and specificity) and no surgical mortality. We defined our primary outcome measure as the difference in LE between the surveillance versus no surveillance strategies.

Sensitivity Analyses

The ranges of parameters explored in the deterministic sensitivity analyses are in Table 1. We varied the length of surveillance from 1 to 10 years. We also changed the risk of developing gallbladder cancer from 0.9 to 1.1 times the age-based values presented in Figure 2. Additionally, we incorporated risks associated with surveillance by varying the ultrasound test performance characteristics for detecting gallbladder cancer and including laparoscopic cholecystectomy surgical mortality. We performed a three-way deterministic sensitivity analysis varying sensitivity (50%−100%), specificity (50%−100%), and surgical mortality rates. We evaluated surgical mortality rates of 0%, 50%, 100%, and 150% of the sex- and age-specific estimates derived from the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) data from the years 2005 to 2019 (e-only Supplemental Table 1) [19]. We assumed that all patients with a positive ultrasound test underwent laparoscopic cholecystectomy and were at risk of surgical mortality.

Table 1.

Parameter estimates and ranges in base-case and sensitivity analyses

Parameter	Base-Case Estimate	Sensitivity Analysis Range Tested
Probability of developing gallbladder cancer	Varied by sex and age [14,17,18]	90%−110% times the base-case value
Gallbladder cancer relative survival	Varied by sex, age,† and time (in years) since diagnosis [14]	Not varied
Mortality from causes other than gallbladder cancer*	Varied by sex, age, and comorbidity level [13,15,16]	Not varied
Length of surveillance	5 years [7]	1–10 years (assumption)
Laparoscopic cholecystectomy surgical mortality rate	0 (assumption)	0%, 50%, 100%, 150% times the sex- and age-specific NSQIP value‡
Sensitivity and specificity of surveillance imaging	1.0 (assumption)	0.5–1.0

NSQIP = National Surgical Quality Improvement Program.

^*More details in the Methods.

^†According to 5-year age groups. More details in the Methods.

^‡Age- and sex-specific surgical mortality risk from the American College of Surgeons National Surgical Quality Improvement Program [19]. See e-only Supplemental Material.

RESULTS

Base-Case Results

All cohorts evaluated had little benefit from 5 years of ultrasound surveillance. LE benefit was lower for men and for patients with severe comorbidities. Women had greater projected LE gains than men due to a higher risk of developing gallbladder cancer and longer LE. The base-case results were mildly sensitive to patient age. Women with no comorbidities aged 66, 70, 75, and 80 gained 1.46, 1.68, 1.70, and 1.45 days, respectively. Men with no comorbidities aged 66, 70, 75, and 80 gained 0.67, 0.80, 0.89, and 0.75 days. Our results were sensitive to patient comorbidity level. For women aged 70 with no, mild, moderate, and severe comorbidities, LE gains were 1.68, 1.45, 1.37, and 1.04 days. For men aged 70 with no, mild, moderate, and severe comorbidities, LE gains were 0.80, 0.75, 0.65, and 0.46 days. Results of the base-case analysis are presented in Figure 3 and Table 2.

Fig. 3.

Projected maximal life expectancy benefit associated with 5 years of ultrasound-based surveillance in (a) women and (b) men with a small, incidental gallbladder polyp: base-case results.

Table 2.

Model-predicted life expectancy in surveillance and no surveillance strategies

Sex and Comorbidity Level	Life Expectancy at Age 66 (y)	Life Expectancy at Age 70 (y)	Life Expectancy at Age 75 (y)	Life Expectancy at Age 80 (y)
Women
No comorbidity
Surveillance	18.768	16.303	13.180	10.224
No surveillance	18.764	16.299	13.175	10.220
Mild comorbidity
Surveillance	17.276	14.463	11.639	9.129
No surveillance	17.272	14.459	11.635	9.126
Moderate comorbidity
Surveillance	15.988	13.849	10.665	8.242
No surveillance	15.984	13.845	10.662	8.239
Severe comorbidity
Surveillance	13.377	11.007	8.543	6.450
No surveillance	13.374	11.004	8.540	6.448
Men
No comorbidity
Surveillance	15.699	13.596	10.915	8.432
No surveillance	15.697	13.594	10.913	8.430
Mild comorbidity
Surveillance	14.836	12.886	10.787	8.015
No surveillance	14.835	12.884	10.784	8.013
Moderate comorbidity
Surveillance	13.682	11.561	9.126	6.903
No surveillance	13.680	11.559	9.124	6.901
Severe comorbidity
Surveillance	10.483	8.840	6.981	5.240
No surveillance	10.482	8.839	6.980	5.239

Sensitivity Analyses

We centered our sensitivity analysis on the cohorts of men and women who could achieve the largest LE benefit through surveillance to expose the impact of varying a given model parameter on our results. In our analysis, these cohorts were 73-year-old women with no comorbidities and 74-year-old men with no comorbidities.

Length of the Surveillance Period.

Our results were sensitive to the length of the surveillance period (Fig. 4). When we extended the surveillance period to 10 years, women aged 73 with no comorbidities gained 3.04 days, compared with 1.72 days in the base-case analysis with 5 years of surveillance. Men aged 74 with no comorbidities gained 1.47 days relative to 0.89 days. Though the LE benefit approximately doubled when we doubled the length of the surveillance period, LE gains were at most about 3 days for all patient cohorts. We saw a similar trend in the opposite direction when we decreased the surveillance period to 1 year.

Fig. 4.

Sensitivity analysis on the length of the surveillance period (1–10 years) on the life expectancy benefit of surveillance of women (a) and men (b) of varying comorbidity level with a small incidental gallbladder polyp. For each comorbidity level, the lighter shade of color indicates that the life expectancy benefit is less than the base-case estimate. In comparison, the darker shade indicates that the life expectancy benefit is greater than the base-case estimate. The lower bound is 1 year, and the upper bound is 10 years. We centered our sensitivity analysis on the cohorts of women and men who could achieve the largest life expectancy benefit through surveillance to best expose the impact of varying a given model parameter on our results.

Gallbladder Cancer Risk.

Our results were mildly sensitive to gallbladder cancer risk (Fig. 5). When we varied the probability of developing gallbladder cancer from 0.9 to 1.1 times the age-specific base-case value, women aged 73 with no comorbidities gained 1.55 to 1.90 days, respectively. Men aged 74 with no comorbidities gained 0.80 to 0.98 days. Overall, LE benefit was slightly greater for patient cohorts with increased cancer risk.

Fig. 5.

Sensitivity analysis on the probability of developing gallbladder cancer (0.9–1.1 times the base-case estimate) on the life expectancy benefit of surveillance of women (a) and men (b) of varying comorbidity level with a small incidental gallbladder polyp. For each comorbidity level, the lighter shade of color indicates that the life expectancy benefit is less than the base-case estimate. In comparison, the darker shade indicates that the life expectancy benefit is greater than the base-case estimate. The lower bound is 0.9 times the base-case estimate, and the upper bound is 1.1 times the base-case estimate. We centered our sensitivity analysis on the cohorts of women and men who could achieve the largest LE benefit through surveillance to best expose the impact of varying a given model parameter on our results.

Sensitivity, Specificity, and Surgical Mortality.

The results of a three-way sensitivity analysis on test sensitivity, specificity, and surgical mortality are presented in Figure 6. LE gains were particularly sensitive to surgical mortality and test specificity. When we assumed no surgical mortality, a lower test specificity resulted in a larger LE gain—this was because, in the absence of harms associated with false-positive test results and cholecystectomy, more patients were spared the lifetime risk of developing gallbladder cancer. However, as we increased the surgical mortality to reflect the ACS NSQIP values [19] while still assuming perfect sensitivity, the threshold at which test specificity resulted in positive LE gains was 96% for women and 99% for men.

Fig. 6.

Gradient maps showing the life expectancy benefits associated with varied test sensitivity and specificity for four sex- and age-specific surgical mortality rates. Each map depicts life expectancy benefits associated with surveillance for a cohort of (a) women aged 73 with no comorbidities and (b) men aged 74 with no comorbidities. See e-only Supplement for information on the surgical mortality estimates from the National Surgical Quality Improvement Program (NSQIP).

DISCUSSION

Using decision-analytic techniques, we found that even with theoretical, overly optimistic assumptions about the performance of ultrasound in identifying gallbladder cancers in the setting of polyp surveillance (6–7 to <10 mm), maximal possible LE gains are minimal. When we considered more realistic assumptions of ultrasound performance and included surgical risks, surveillance of small gallbladder polyps did more harm than good. Perfect surveillance over the 5-year period supported by 2020 Canadian recommendations [7] was projected to result in LE gains of less than 2 days for women and under 1 day for men, across all age groups evaluated, even in individuals with no comorbidities. Reducing surveillance to 1 year, as recommended by the SRU for many small polyps [8], resulted in lower LE gains, as shown in Figure 4. Given the extremely modest projected benefits, despite unrealistic assumptions of perfect cancer detection and treatment, our findings support reduction or elimination of surveillance in many patients with small, incidentally detected polyps.

Our findings are driven by the low prevalence of gallbladder cancer among individuals with polyps [6]. Because of this low observed prevalence, even if there were no harm incurred from gallbladder resection, maximum benefits would still be minimal. Moreover, when even slightly imperfect assumptions of cancer detection and treatment are considered—including slightly imperfect test specificity and low mortality risks from laparoscopic cholecystectomy—projected benefits of surveillance disappear. Our findings under these more realistic assumptions are sobering and bring to light the possibility that, although well-intended, extended surveillance regimens for incidentally detected gallbladder cancers could cause more harm than benefit.

The imaging literature on the management of gallbladder polyps, to date, has principally focused on the determination of features, particularly size and patient characteristics, that are associated with the greatest cancer risk [3,9,10,20,29]. More recent guidance from the SRU and European consensus recommendations has additionally incorporated polyp morphology [8,29]. Existing recommendations are instrumental in reducing the overall number of patients who undergo unnecessary surveillance—specifically, by eliminating follow-up for very small polyps (<6–7 mm) and triaging patients who are at higher risk based on patient characteristics, larger polyp size, and advanced morphology to surgery. However, when considering patient-level decision-making concerning surveillance, additional factors warrant consideration, including patients’ cancer risk relative to those without polyps, their age and comorbidities, and risks associated with treatment. Decision-analytic models, such as those used in the current analysis, enable multiple health risks and benefits to be mathematically weighed to determine the net benefits (or risks) incurred from a given surveillance intervention. In this way, decision-analytic models provide important, complementary insights to studies that directly analyze imaging data.

The current analysis, as well as similar analyses conducted by our group and others concerning the benefits and risks of surveillance of incidental findings in various patient groups [11,12,21], underscores the importance of developing more rigorous criteria for patients’ participation in imaging surveillance. In the context of an incidental finding, if the risk of treatment outweighs the risk of cancer, then encouraging surveillance would involve both high-risk and high-cost care. However, determining and weighing such risks with confidence is difficult, because they vary from patient to patient. Moreover, primary care physicians understandably rely on radiologists’ determination of a given lesion’s risk—in many scenarios, this risk is small but real, and follow-up seems prudent. However, the responsibility for any further weighing of risks is displaced to the primary care physician, who would need to overturn the radiologists’ recommendation, if they believe that the patients’ risk of harm from surveillance and treatment was greater than the cancer risk associated with the lesion in question. This puts the primary care physician in a very difficult position. Ideally, radiologists’ recommendations should incorporate, at minimum, language that accommodates the validity of not following up low-risk lesions under certain conditions. Radiologist recommendations that are more tailored to individual patients would be even better. New SRU and European consensus recommendations support shorter follow-up for most gallbladder polyps under surveillance and also account for polyps’ morphologic features, in addition to patient characteristics [8,29].

Importantly, debate concerning the appropriateness of different recommendations for gallbladder polyp surveillance is well known among radiologists. Notably, results of a recent survey of the Fellows of the SRU indicated that a paradigm shift toward higher size thresholds for triggering follow-up—and further reductions in follow-up recommendations—would be favorable among subspecialized radiologists [10]. Two recent observational studies [22,23] suggested that the gallbladder polyp size thresholds for referral to surgery, when aligning with guidance from the 2017 European and 2020 Canadian consensus recommendations [3,7], likely resulted in the overtreatment of small gallbladder polyps. Given how common upper abdominal ultrasound examinations are in both academic and private practice settings, the conundrum of how to best manage gallbladder polyps affects both settings—and subspecialists and generalists alike—on a daily basis.

Our study has limitations that warrant mention. First, decision-analytic models represent simplifications of the natural history of disease processes, as well as of the interventions assessed. In the current analysis, of particular note, we did not incorporate different surveillance intervals (eg, every 6 months or every year). In so doing, projections of benefits were unrealistically high, as we did not incorporate the development and progression of interval cancers, occurring between surveillance examinations, during the surveillance period. Second, data on the natural history of gallbladder polyps—particularly in the size range of interest—remain limited. The current analysis focuses on assumptions from the largest study available, but cancer risk assumptions were not specific to the polyp size group of interest. Even so, when increasing cancer risk in sensitivity analysis, the effects on our findings were minimal. Third, we incorporated patient sex, age, and comorbidity level as risk factors for gallbladder cancer; however, due to limitations in data availability, we have not incorporated polyp morphology [24,25], ethnicity and geographic location [26–28]. Fourth, in sensitivity analysis, we incorporated the risks of laparoscopic cholecystectomy from ACS NSQIP. However, in the setting of gallbladder carcinoma, additional surgical intervention would likely be necessary. As such, the incorporated surgical mortality risks may be underestimated, leading to overestimated net benefits from surgical intervention in our analysis. Again, even with optimistic assumptions, the projected benefits were minimal. Fifth, a formal cost-effectiveness analysis was beyond the scope of our current study. However, such analysis would provide specific insight into the value of varied management recommendations for imaging surveillance, from economic and societal perspectives.

In conclusion, using decision-analytic modeling methods, we found that projected LE gains associated with imaging-based gallbladder surveillance for individuals ages 66 or older with small polyps (6–7 to <10 mm) were modest, totaling 3 days or less in all scenarios evaluated. With increased patient comorbidities—and with considerations of imperfect test performance during surveillance and risks of surgical mortality—projected benefits further diminished or disappeared. Our findings indicate the need for continued re-evaluation of current recommendations regarding gallbladder polyp surveillance, particularly for individuals with limited LE, and for consideration of more patient-specific guidance.

TAKE-HOME POINTS.

• Current recommendations from national and international organizations provide varying guidance for the management of small gallbladder polyps, with most recommendations for small polyps (6–7 to <10 mm) ranging from <1 year to 5 years of surveillance.

• Projected LE gains associated with gallbladder surveillance for individuals ages 66 or older with small polyps were modest, totaling less than 3 days in key scenarios evaluated.

• With increased comorbidities, projected benefits further diminished.

• Ongoing re-evaluation of current recommendations regarding gallbladder polyp surveillance is likely warranted, particularly for individuals with limited LE.