Development and Validation of a Symptom Index for Advanced Hepatobiliary and Pancreatic Cancers: The NCCN-FACT Hepatobiliary-Pancreatic Symptom Index (NFHSI)

Abstract

Objective

The 45-item Functional Assessment of Cancer Therapy – Hepatobiliary (FACT-Hep) questionnaire assesses health-related quality of life in patients with liver, bile duct and pancreatic cancers. Although the FACT-Hep was initially derived from patient input, we sought to verify adequate coverage of items by soliciting open-ended input from patients with advanced disease.

Methods

As part of a larger study in collaboration with the NCCN, 50 people (60% male, 80% Caucasian, average age 60.4 yrs) with Stage 3 or 4 hepatobiliary or pancreatic cancer were recruited. Participants generated and ranked up to 10 important symptoms and concerns that physicians should monitor when assessing the value of chemotherapy. Patients were also able to provide open-ended, qualitative information that was evaluated systematically. Ten expert physicians also provided input on priority symptoms.

Results

The resulting 18-item NCCN-FHSI (NFHSI-18) demonstrated high internal consistency (α = 0.89) and moderate to strong correlations with measures of physical well-being (ρ = 0.76), emotional well-being (ρ = 0.52), and functional well-being (ρ = 0.57). Scores on the NFHSI-18 were also highly correlated with the original hepatobiliary scale of the FACT-Hep (ρ = 0.82; all p<0.001). Compared to patients with better performance status, patients with poor performance status had worse NFHSI-18 symptom scores, F (3, 47) = 9.74; p=0.0003.

Conclusions

The NFHSI 18 assesses symptoms of importance to patients with hepatobiliary and pancreatic cancers and-demonstrates promising measurement properties. The scale is a good candidate for brief symptom assessment in clinical trials.

Hepatobiliary cancer is a broad term that refers to primary malignancies of the liver, bile ducts, and gallbladder, with the most common malignancies being hepatocellular carcinoma, cholangiocarcinoma, and gallbladder cancer. Pancreatic adenocarcinomas are the second most common of the cancers of the digestive system. Together, hepatobiliary and pancreatic cancers are a highly lethal group of cancers for which few effective treatment options exist.^{1, 2} In 2012, there will be an estimated 28,720 new cases of liver and intrahepatic bile duct cancers, 9810 new cases of gall bladder cancer, and 43,920 new cases of pancreatic cancer. As a group, hepatobiliary and pancreatic cancers are estimated to account for accounted for over 61,000 cancer deaths a year.³

Hepatocellular carcinoma (HCC) can develop as a result of viral infections caused by hepatitis B virus and/or hepatitis C virus or via non-viral routes, such as alcoholic cirrhosis.^{1, 4} The prevalence has been increasing in recent years and it is currently the third leading cause of cancer deaths worldwide and the ninth leading cause of cancer death in the United States.⁵ The most recent estimates of worldwide incidence vary from >20 in 100,000 in areas with endemic hepatitis B viral infection to <5 in 100,000.⁶ Published guidelines exist for the surveillance of patients for HCC, however compliance with screening regimens is highly variable, and HCC can often present at advanced stages.^7-9 Treatment options for early disease include liver transplantation, locoregional therapy, or surgical resection.^{1, 7} By contrast, treatment options are limited with advanced stages of HCC, with sorafenib being the most widely used systemic treatment.¹⁰

Cholangiocarcinoma is a primary malignancy of the epithelium of the bile ducts that is also rare and difficult to treat.¹¹ The locations of the two main subtypes of cholangiocarcinoma are in the intrahepatic bile ducts or the extrahepatic bile ducts, which also includes the confluence of the right and left bile ducts (hilum).¹¹ Aside from primary sclerosing cholangitis or choledochal cystic disease, there are no established risk factors or surveillance guidelines for cholangiocarcinoma.¹ The only available curative therapies are surgical resection or liver transplantation, however only a minority of patients present with curable disease.^{11, 12} There is no established role for chemotherapy or radiation therapy in prolonging survival for patients with unresectable disease.¹³

Gallbladder cancer is a rare but lethal hepatobiliary cancer that can arise in the setting of chronic or recurrent cholecystitis.¹ However it is estimated only 0.5-1.5% of patients with potential risk factors develop disease, thus no surveillance guidelines exist.¹⁴ Often, gallbladder cancer is found incidentally during surgery for another indication.^{1, 14} It frequently presents at advanced stages because it is a highly aggressive tumor. The only curative treatment for local disease is radical cholecystectomy, but this is only possible in 10-30% of patients, and recurrence rates are high.^15-17 Palliative chemotherapy or radiation therapy may be of benefit for more advanced disease, although no definitive treatment trials have been conducted to date.^{16, 18}

Pancreatic adenocarcinomas arise from the pancreatic ductal epithelium. There are few established risk factors for the development of pancreatic cancer, and no screening guidelines for these malignancies exist.² Patients with pancreatic cancer classically present with painless jaundice, which signifies invasion into the biliary tree, and thus there are few warning signs until advanced disease is present.¹⁹ Surgical resection is the primary treatment option in early pancreatic cancer.^{20, 21} In more advanced disease, systemic chemotherapy is an option, however survival rates remain very low.^{20, 22}

As a group, the hepatobiliary and pancreatic cancers have quite low 5-year survival rates and therefore palliative symptom management of advanced cancers is of importance. It is therefore surprising that there have been relatively few studies investigating health related quality of life (HRQOL) in patients with these cancers. Sun et al. prospectively studied HRQOL in 45 patients with hepatobiliary malignancies and found that their HRQOL scores, as measured by the Functional Assessment of Cancer Therapy – Hepatobiliary (FACT-Hep) ²³, were lower (i.e. worse) at baseline compared to normative samples and worsened over the three months the patients were followed.²⁴ Steel et al. studied HRQOL in 81 patients with HCC, also using the FACT-Hep, and found that HRQOL was significantly worse than a sample of patients with chronic liver disease and the general population.²⁵ Based upon these studies, it is apparent that HRQOL is diminished in patients with hepatobiliary cancers.

Cella and colleagues²⁶ identified priority symptom and quality of life concerns in several advanced cancers, including hepatobiliary and pancreatic cancers. Using questions from the Functional Assessment of Cancer Therapy (FACT) quality of life measurement system, they polled expert physicians and nurses (N=455) at 17 National Comprehensive Cancer Network (NCCN) member institutions to derive a set of brief, clinically relevant symptom indexes for nine tumor sites (i.e. bladder, brain, breast, colorectal, head and neck, hepatobiliary/pancreas, lung, ovarian, and prostate). As part of the survey, each expert clinician iteratively rated the five most important symptoms to assess in the evaluation of drug treatment for advanced disease. Several symptom indices, including the FACT – Hepatobiliary Symptom Index (FHSI-8), were developed as part of that process.

Although the FACT-Hep and FHSI-8 were derived with prior patient input, we sought to verify adequate coverage of items by soliciting open-ended input from a new sample of advanced hepatobiliary and pancreatic cancer patients, naïve to the content of these questionnaires. In this paper, we describe development of a brief symptom index comprised of the most important symptoms, as defined by patients and expert clinicians. Use of such a scale in future clinical trials and observational studies may provide relatively brief, yet valuable input on symptom burden.

Method

Patient Eligibility and Recruitment

Data for the present study come from a larger study¹² which sought to develop a series of patient-centered, tumor-specific HRQOL indexes. Our focus in the present analyses was on patients’ and providers’ experience with advanced hepatobiliary and pancreatic cancer.

Patients were eligible if they were at least 18 years of age and had stage III or IV hepatobiliary or pancreatic cancer. Patients could not have any other primary malignancy in the previous five years, except non-melanoma skin cancer. In addition, patients must have had at least two cycles of chemotherapy (or one month of treatment with non-cyclical chemotherapy).

Patients with hepatobiliary and pancreatic cancers were recruited from a subset of NCCN member institutions and community support agencies. Specifically, participants from Duke University Medical Center (Durham, NC, n=15), Fred Hutchinson Cancer Research Center (Seattle, WA, n=15), the Robert H. Lurie Comprehensive Cancer Center of Northwestern University Feinberg School of Medicine (Chicago, IL, n=19), and a community-based support agency in Chicago (n=1). The study protocol was approved by local institutional review boards, and all patients provided informed consent.

Physician Eligibility and Recruitment

Physician input was solicited to assess which symptoms were most likely to be disease-related versus treatment-related. Eligible physician participants were currently in practice and had at least three years’ experience treating a minimum of 100 patients with hepatobiliary and/or pancreatic cancer. Physicians were recruited by e-mail from 20 NCCN member institutions by NCCN staff or by study staff at the participating institutions. Ten physician-experts provided input on the symptom index.

Procedures

In an open-response interview format, patients were first asked: “Please think of the full range of your experience receiving drug treatment for your illness. Please tell me what you think are the most important symptoms or concerns to monitor when assessing the value of drug treatment for your illness.” As a follow-up question, participants were asked: “Please tell me on a scale of 0-10 (with 0 being not important and 10 being extremely important) how important each symptom or concern is to you.” The open-response format allowed for patient input without potential priming effects of a pre-constructed symptom list.

Next, the 50 patients with hepatobiliary or pancreatic cancer were asked to review a symptom/concern checklist that contained 37 items nominated by the physician experts and from the FACT-Hep questionnaire. The checklist was identical in structure to that administered to physicians and nurses in our previous study.²⁷ To control for response bias due to order effect, four versions of the checklist were administered. Patients were first asked to select no more than 10 symptoms or concerns that they felt were “the most important symptoms or concerns to monitor when assessing the value of drug treatment for advanced hepatobiliary/pancreatic cancer.” Patients were then asked to select up to five of the top 10 ranked concerns as “the very most important.” Space was provided for respondents to write in symptoms or concerns that were not listed. To help determine which items would be included in the final symptom index, the frequency of chance endorsement was calculated by dividing 5 (the final number of “very most important symptoms”) by 37 (the total number of items in the hepatobiliary cancer-specific checklist) and then multiplying this number by the number of participants (i.e. 50). This a priori criterion, along with other considerations, such as items recommended by hepatobiliary/pancreatic cancer experts,²⁸ was used to recommend the final symptom index.

After providing their symptom rankings, patients in this study were asked to complete the FACT-Hep,²³ which contains the 27-item Functional Assessment of Cancer Therapy-General (FACT-G)²⁹ questionnaire as well as the 18-item Hepatobiliary Cancer Subscale.

Eligible physicians (n=10) were directed to a web-based survey where they were asked to rate all symptoms on a 5-point scale based on the degree to which the item was disease-related versus a treatment side-effect. Physicians were also given the opportunity to rate symptoms as neither disease- nor treatment-related.

Results

Sample

Table 1 summarizes the sociodemographic and clinical status of the 50 enrolled participants. Patients had a median age of 58.5 years (range 30 – 87 years). The sample was 40% female and was predominately Caucasian (80%) and well educated (46% with bachelor's degree or more). Half the patients reported at least some symptoms (adapted ECOG performance status rating, or PSR=1) and over a third reported symptoms severe enough to require at least some bed rest during their waking day (PSR=2; 3).

Table 1.

Sociodemographic and performance status of sample (n = 50)

	M (SD)
Age in years
M (SD)	60.5 (12.6)
Median (range)	58.5 (30-87)

	n	%
Female	20	40
Race/Ethnicity
White	40	80
African-American	6	12
Asian	3	6
Other	1	2
Education
< 12th grade	3	6
HS diploma or GED	9	18
Vocational school or some college	15	30
Bachelor's	18	36
Professional or graduate degree	5	10
Current Occupational Status
Homemaker	3	6
Unemployed	1	2
Retired	20	40
On Disability	12	24
On Leave of Absence	3	6
Employed, Full-time	8	16
Employed, Part-time	3	6
Patient-rated ECOG Performance Status Rating
0, normal activities without symptoms	7	14
1, some symptoms, but do not require bed rest during the waking day	25	50
2, require bed rest for less than 50% of waking day	15	30
3, require bed rest for more than 50% of waking day	3	6

Table 2 identifies whether the 18 items on the new NCCN-FACT Hepatobiliary-Pancreatic Symptom Index (NFHSI-18) were rated as a “Top 5” symptom at greater than chance levels by patients. We also present a summary of whether symptoms were highly rated by experts in our original paper.³⁰ As seen from the ratings, some items, such as fatigue/lack of energy, nausea, and discomfort/pain in abdomen were highly ranked by both groups. Table 2 also presents a comparison of the NFHSI-18, the FACT-Hep²³ and the original FHSI-8.³⁰

Table 2.

Items comprising the NCCN-FACT Hepatobiliary-Pancreatic Symptom Index (NFHSI)

NCCN-FACT Hepatobiliary Symptom Index Items		Endorsed as a “Top 5” symptom at greater than chance levels by patients?	Endorsed as a “Top 5” symptom at greater than chance levels by experts?	On FACT-Hep?*	On FHSI-8?
Label	Item Content
GP1	I have a lack of energy	Yes	Yes	Yes	Yes
GP4	I have pain	Yes	Yes	Yes	Yes
C2	I am losing weight	No	Yes	Yes	Yes
HI7	I feel fatigued	Yes	Yes	Yes	Yes
CNS7	I have pain in my back	No	Yes	Yes	Yes
Hep2	I am bothered by jaundice or yellow color to my skin	No	Yes	Yes	Yes
GP6	I feel ill	Yes	No	Yes
Hep8	I have discomfort or pain in my stomach area	Yes	Yes	Yes	Yes
GP2	I have nausea	Yes	Yes	Yes	Yes
GP3	Because of my physical condition, I have trouble meeting the needs of my family	Yes	No	Yes
GE6	I worry that my condition will get worse	Yes	No	Yes
GE1	I feel sad	Yes	No	Yes
C6	I have a good appetite	Yes	No	Yes
GF5	I am sleeping well	Yes	No	Yes
GP5	I am bothered by the side effects of treatment	---	---	Yes
AN7	I am able to do my usual activities	Yes	No	Yes
GF3	I am able to enjoy life	Yes	No	Yes
GF7	I am content with the quality of my life right now	Yes	No	Yes

^*The FACT-Hep Total scale includes an additional 27 items not listed here. Expert ratings come from Yount et al (2002)

Physician ratings of the NFHSI-18 items are summarized in Table 3. Most symptoms were rated as having at least some relation to the disease itself. Jaundice, pain, and worry that the condition may worsen were more likely than other symptoms to be rated “exclusively disease-related.” Fatigue, ability to do usual activities, nausea, and ability to enjoy life were most likely to be rated “too close to determine” on the disease- versus treatment-related continuum. These ratings were used to help define Disease-Related Symptoms and Functional Well-Being subscales, as well as a single-item Treatment Side Effects indicator.

Table 3.

Physician ratings of NFHSI-18 symptoms

NCCN-FACT Hepatobiliary-Pancreatic Symptom Index Items	Exclusively Disease-Related	Predominately Disease-Related	Too Close to Determine	Predominately Treatment-Related	Exclusively Treatment-Related	Neither Disease-or Treatment-related
I have a lack of energy		80%	20%
I have pain	50%	50%
I am losing weight	20%	70%	10%
I feel fatigued		30%	60%			10%
I have pain in my back	40%	60%
I am bothered by jaundice or yellow color to my skin	90%	10%
I feel ill		80%	20%
I have discomfort or pain in my stomach area	30%	50%	10%	10%
I have nausea		50%	40%	10%
Because of my physical condition, I have trouble meeting the needs of my family		60%	30%			10%
I worry that my condition will get worse	50%	50%
I feel sad	10%	80%	10%
I have a good appetite		40%	40%	10%		10%
I am sleeping well		50%	40%	10%
I am bothered by the side effects of treatment
I am able to do my usual activities		40%	60%
I am able to enjoy life		50%	40%	10%
I am content with the quality of my life right now		70%	20%	10%

As seen in Table 4, the new NFHSI-18 and its subscales demonstrated good internal consistency reliability, with α > 0.82. Scores on the FACT-G, its subscales, and the Hepatobiliary Cancer Subscale from the FACT-Hep were at expected levels for this patient population.

Table 4.

Descriptive statistics of NFHSI-18 and other scales (n=50)

	α	Possible Range	M (SD)
NFHSI-18 Total	0.89	0 – 72	45.7 (12.8)
NFHSI-DRS-14 (Disease Related Symptoms)	0.87	0 – 56	36.8 (10.2)
NFHSI-FWB-3 (Functional Well-Being)	0.82	0 – 12	6.4 (3.1)
NFHSI-TSE-1 (Treatment Side Effects)	N/A*	0 – 4	2.5 (1.2)
FACT-G Total		0 – 108	72.7 (16.8)
FACT-G Physical Well-being		0 – 28	18.9 (6.4)
FACT-G Social Well-being		0 – 28	22.0 (5.7)
FACT-G Emotional Well-being		0 – 24	16.1 (4.9)
FACT-G Functional Well-being		0 – 28	15.8 (6.4)
FACT-Hep Total		0 – 180	123.8 (24.8)
FACT-Hep Hepatobiliary Cancer Subscale		0 – 72	51.1 (9.6)

Notes: Alpha not calculated for the single-item NFHSI-TSE-1

We also examined the association of scores on the NFHSI-18 scales with the FACT-G, FACT-Hep, and its subscales. Overall, the inter-correlations are moderate to large in magnitude in the expected direction. After removal of overlapping items, the total NFHSI-18 was most highly correlated with the FACT-G Physical Well-being (ρ=0.76), FACT-G Functional Well-being (ρ=0.57), FACT-G Emotional Well-being (ρ=0.52), and the Hepatobiliary Cancer Subscales (ρ=0.82, all p < 0.001).

Table 5 presents validity data for the NFHSI-18 scales, based on comparison of scores from ECOG sub-groups. As expected, scores on the NFHSI-18 scales were worse for patients who reported more functional impairment on the ECOG performance status rating. The ANOVAs for these comparisons were all statistically significant (p < 0.05).

Table 5.

NFHSI scores [M (SD)] as a function of patient-reported ECOG performance status rating

	n	NFHSI-18*	NFHSI-DRS**	NFHSI-FWB***	NFHSI-TSE****
0, normal activities without symptoms	7	56.3 (8.7)	44.4 (6.2)	9.0 (2.2)	2.9 (1.2)
1, some symptoms, but do not require bed rest during the waking day	25	48.9 (10.7)	39.7 (9.2)	6.4 (2.8)	2.8 (1.1)
2, require bed rest for less than 50% of waking day (n=15) and 3, require bed rest for more than 50% of the waking day (n=3)	18	37.2 (11.9)	29.8 (9.0)	5.6 (3.3)	1.9 (1.1)

^*F(2, 47) = 9.74, p = 0.0003

^**F(2,47) = 9.79, p = 0.0003

^***F (2, 47) = 3.51, p = 0.04

^****F (2, 47) = 3.95, p = 0.03

Discussion

In this paper, we present our development process for the 18-item NCCN-Functional Assessment of Cancer Therapy – Hepatobiliary-Pancreatic Symptom Index (NFHSI-18) and its subscales, as well as initial data on the measures’ reliability and validity. The NFHSI-18 scales are moderately to highly correlated with other measures of general quality of life (e.g. FACT-G subscales) and health-related quality of life in patients with hepatobiliary/ pancreatic cancer (e.g. FACT-Hep subscales). Scores on the NFHSI-18 track well with performance status; as ECOG status worsens, NFHSI-18 scores become significantly lower. Based on these early data, we believe the NFHSI-18 represents an improvement on the FHSI-8 and its component subscales may serve as useful endpoints in future clinical trials among patients with advanced hepatobiliary and pancreatic cancer.

With input from patients and physicians, we originally developed the FACT-Hep questionnaire as a broad, comprehensive measure of general and disease-specific symptoms and other aspects of health-related quality of life among patients with hepatobiliary and pancreatic cancers.²³ In an effort to produce a brief, clinically meaningful symptom index, FACT-Hep items were reviewed by 95 clinical experts in hepatobiliary and pancreatic cancers to produce the 8-item FACT Hepatobiliary Symptom Index (FHSI-8). While use of these scales has grown, we recently sought to ensure adequate patient input in brief symptom indexes across 11 tumor types.²⁷

In 2009, the United States Food and Drug Administration (FDA) issued guidance for the pharmaceutical industry that describes how the FDA evaluates PRO instruments used as effectiveness endpoints in clinical trials.³¹ In this document, the FDA described their perspective on how industry sponsors can develop and use study results measured by PRO instruments to support claims in approved product labeling. In clinical trials, a PRO may be used to measure the impact of an intervention on one or multiple aspects of patients’ health status. This could range from a specific symptom response (e.g., pain relief) to more general concepts such as health-related quality of life. The main focus of the FDA guidance is on the evidentiary support for claims of PRO benefit, not general health-related quality of life research.³²

Our development of the NFHSI-18 uses items drawn from the FACIT system and methods consistent with the recent FDA guidance. The end product is a symptom index specific to advanced hepatobiliary and pancreatic cancers, reflecting the values and language of patients themselves. The lion's share of items in the symptom index were judged by experts to be mostly related to the disease itself and are expected to improve with effective treatment. Symptoms that may be treatment side effects are expected to vary as new therapies become available. Beyond the clinical value of such indexes, scales like the NFHSI-18 may also contribute significantly to satisfying regulatory requirements for a standardized tool to evaluate drug efficacy with respect to relevant symptoms. Further, by aligning the perspectives of physicians, nurses, and patients diagnosed with these advanced cancers, the most current set of symptom indexes will represent a comprehensive list of the symptoms prioritized by the most integral stakeholders involved in patient care.

Our sample size is appropriate for evaluation of an adapted patient-reported outcome; however, additional psychometric studies in larger samples of patients will be useful to assess the generalizability of our findings. Specifically, future prospective studies of patients with hepatobiliary and pancreatic cancers will provide useful data to evaluate the scale's responsiveness to clinically significant change over time. In a separate report,²⁸ we evaluated NFHSI-18 data collected as part of a randomized controlled Phase II study evaluating efficacy and safety of conatumumab, ganitumab or placebo combined with gemcitabine in approximately 100 patients with metastatic pancreatic cancer. Reliability, validity, and responsiveness to change were quite promising and in line with findings from the current study.

One clear advantage of the FACT hepatobiliary measures is that they are complete subsets of one another. All items of the original FHSI-8 are included in the NFHSI-18, which in turn are all included in the FACT-Hep. This allows researchers the flexibility to administer the scale that fits the specific assessment needs at hand. It also allows comparison of results from earlier trials with ones that use any of the more abbreviated FACT scales. Selection of a specific scale should be guided by several factors, including purpose and constraints of assessment and stage of disease.