Association between nociplastic pain and premature endocrine therapy discontinuation in breast cancer patients

Abstract

Purpose:

At least five years of adjuvant endocrine therapy (ET) is recommended for patients with hormone receptor-positive invasive breast cancer to reduce cancer recurrence risk. Up to half of patients prematurely discontinue ET, often due to musculoskeletal pain. Nociplastic pain is abnormal central nervous system pain processing without evidence of tissue or neuronal damage. This study aimed to evaluate the relationship between baseline nociplastic pain and ET discontinuation.

Methods:

This was a retrospective, single center, cohort study. Included patients were female, had stage 0-III invasive breast cancer, did not receive neoadjuvant therapy, and completed quality of life questionnaires prior to breast surgery, including Fibromyalgia Survey for nociplastic pain. Clinical data including duration of ET were abstracted from the medical record. Patient characteristics were analyzed with t-tests and chi-squared tests, as appropriate. Univariate and multivariable regressions were performed with Cox proportional hazard models.

Results:

681 patients diagnosed between 2012–2019 met inclusion criteria; 480 initiated ET and were included in the analysis. Of these 480 patients, 203 (42.3%) prematurely discontinued initial ET therapy. On univariate analysis, tamoxifen use (hazard ratio [HR] 0.70, p=0.021) and premenopausal status (HR 0.73, p=0.04) were inversely associated with ET discontinuation, while Fibromyalgia Score was positively associated (HR 1.04, p=0.043). On multivariable analysis, baseline Fibromyalgia Score remained associated with ET discontinuation.

Conclusion:

Nociplastic pain present prior to surgery was associated with premature ET discontinuation. Fibromyalgia Score screening may be useful for evaluating ET discontinuation risk. Treatments targeting nociplastic pain may be more effective for treating ET-emergent pain.

Introduction:

Adjuvant endocrine therapy (ET) is recommended for at least five years after diagnosis with hormone receptor-positive invasive breast cancer to reduce risk of breast cancer recurrence and mortality [1]. Despite the potential benefit, up to half of patients discontinue ET prior to the recommended five years [2]. One of the most common reasons for discontinuation is toxicity, including hot flashes, vaginal bleeding, fatigue, musculoskeletal pain, and fractures [2–4]. Up to 50% of patients who take aromatase inhibitors (AI) report aromatase inhibitor-associated musculoskeletal symptoms (AIMSS) and up to 30% of patients who take tamoxifen report fatigue or vasomotor symptoms [2, 4]. Although estrogen-deprivation and inflammation have been suggested to contribute to the pathogenesis of ET side effects, the exact mechanism by which these side effects occur remains poorly understood, making them difficult to predict and treat [5].

Historically, pain has been categorized as nociceptive or neuropathic. Nociceptive pain is related to tissue damage, while neuropathic pain refers to pain arising from damage or dysfunction in the nervous system [6]. However, neither definition adequately explains the experience of patients with chronic pain conditions such as fibromyalgia, irritable bowel syndrome, vulvodynia, and temporomandibular disorders, which lack obvious tissue or nerve damage and do not typically respond to anti-inflammatories or opioids [7, 8]. As with AIMSS, estrogen has also been proposed in the pathogenesis of chronic pain. Research demonstrates increased pain exacerbation severity during low estrogen states in premenopausal women with temporomandibular disorders, development of chronic musculoskeletal pain around the time of menopause, and increased severity of baseline pain and pain exacerbations in postmenopausal women with rheumatoid arthritis [5, 9, 10]. Estrogens have been implicated in influencing a number of neurotransmitters, vasodilators, and other neurobiologic substances, but the exact relationship between estrogens and chronic pain conditions remains unclear [11].

In 2016, the International Association for the Study of Pain (IASP) proposed a new category of pain, called “nociplastic” or centralized pain [12]. Nociplastic pain is distinct from nociceptive and neuropathic pain in that it arises from altered nociception in the absence of tissue or neuronal damage; it is thought to be due to abnormal pain processing in the central nervous system [6]. It can be identified using patient-reported outcome questionnaires such as the Fibromyalgia Survey Score, which assesses localized versus widespread pain as well as associated symptoms including fatigue, cognitive dysfunction, and depression [13].

Based on the observations that chronic pain may be increased in low estrogen states, this study investigated whether nociplastic pain present prior to breast cancer treatment is associated with premature ET discontinuation [9, 14]. Our hypothesis was that preexisting nociplastic pain is associated with premature ET discontinuation. We hypothesized this would be true for aromatase inhibitor (AI) therapy, given that AIs act by lowering circulating estrogen levels. The secondary aim was to identify and evaluate other variables for an association with premature ET discontinuation.

Methods:

Study population:

This was a retrospective, single-center, cohort study of patients diagnosed with breast cancer between 2012–2019 who were enrolled in the Michigan Genomics Initiative (MGI) study [15]. Patients met inclusion criteria if they were female, had stage 0-III invasive breast cancer, had not received neoadjuvant chemotherapy or endocrine therapy, and were enrolled in MGI (Online Resource 1).

MGI is an ongoing prospective study consisting of >80,000 patients at the University of Michigan. MGI’s goal is to compile patient genomic, laboratory, and electronic health record (EHR) data to facilitate new medical advances. For this study, participants were prospectively recruited at the time of breast cancer surgery in the preoperative waiting area. Participants provided written informed consent, then completed health-related quality of life (HRQOL) questionnaires and provided research specimens before undergoing surgery. MGI was approved by the Institutional Review Board at the University of Michigan (IRBMED; HUM00028683), and this exploratory analysis was granted an exemption by IRBMED (HUM00169453).

Patient-reported outcomes:

As described, patients completed HRQOL questionnaires immediately prior to undergoing surgery for their breast cancer. Over time, the questionnaire battery included in the MGI project changed. Therefore, only questionnaires that at least 75% of study participants had responded to were included in this analysis.

The Fibromyalgia Survey (FS) was used to assess nociplastic pain with questions about widespread pain and somatic symptom severity [13, 16]. This validated questionnaire has been repeatedly demonstrated in the literature to be positively correlated with symptoms and poor response to treatment across a variety of chronic pain conditions [17–20]. Widespread pain was identified using the Michigan Body Map, a map with 35 body areas on which patients indicated locations where they have experienced pain for longer than three months (scored 0–19). The Symptom Severity Index (SSI) consisted of six questions evaluating fatigue, difficulty thinking, waking up tired, abdominal cramps, headache, and depression (scored 0–12). The sum of these two measures created the FS, evaluated on a 0–31 scale. Higher FS scores reflect more nociplastic pain, and FS ≥ 13 meets criteria for fibromyalgia.

The Brief Pain Inventory (BPI) was used to assess overall worst pain and pain at the surgical site during the week prior to completion of the questionnaire, on a 0–10 scale, with 0 being no pain and 10 being pain as bad as you can imagine [16, 21].

Life satisfaction was a single question measuring general satisfaction with life on a 0–10 scale, with 0 being the least satisfied and 10 being the most satisfied.

Study Outcomes:

Patient characteristics and comorbidities were abstracted from the electronic health record (EHR). Use and discontinuation of ET and reasons for discontinuation were abstracted from oncology provider clinic notes. ET discontinuation was defined as time to discontinuation of first ET agent. For those patients who initiated ET, we calculated the simple proportion of patients who discontinued initial ET before completing five years of therapy and the proportion of patients who discontinued their initial ET due to side effects, ignoring censoring. We also evaluated variables (age, BMI [body mass index], ET drug class [tamoxifen, AI], prior chemotherapy [yes, no], radiation therapy [yes, no], patient-reported outcomes [FS, BPI worst pain, life satisfaction]) potentially associated with premature discontinuation of initial ET.

Statistical Methods:

Patient characteristics were compared between different patient groupings based on hormone-receptor and ET status using t-tests and chi-squared tests, as appropriate. Univariate and multivariable regressions on those who initiated ET were performed with Cox proportional hazard models. There was evidence of collinearity between ET and menopausal status as well as violation of the proportional hazard assumption for menopausal status, so the multivariable model excluded this variable. P values <0.05 were considered statistically significant. All statistical analyses were performed using R version 4.1.0.

Results:

Patient characteristics:

A total of 681 participants met inclusion criteria. The average age was 57 years (standard deviation [SD] 12) and average BMI was 29 (SD 7) (Table 1). Most patients were postmenopausal (68%) and white (89%). A total of 231 (34%) received adjuvant chemotherapy, and 397 (61%) received adjuvant radiation.

Table 1:

Baseline characteristics of the total cohort.

	Total (N = 681)
Age, mean (SD)	57 (12)
BMI, mean (SD)	29 (7)
Menopausal status
Postmenopausal	451 (68%)
Premenopausal	213 (31%)
Missing	17 (2.5%)
Race
White	603 (89%)
Black	33 (4.9%)
Asian	14 (2.1%)
Other	30 (4.4%)
Missing	1 (0.1%)
Received adjuvant chemotherapy	231 (34%)
Received adjuvant radiation therapy	397 (61%)
Receptor status
ER positive	606 (89%)
PR positive	534 (79%)
HER2 positive	82 (12%)

BMI: body mass index; ET: endocrine therapy: SD: standard deviation.

Of the 681 participants in the study, 613 had hormone receptor-positive disease and 503 of those (82%) initiated ET (Table 2). Compared to those who initiated ET, those who did not initiate ET were more likely to have negative lymph nodes or nodes that were not assessed (no ET: 91 [83.7%] vs ET: 357 [71.0%], p=0.023), less likely to have undergone adjuvant radiation (38 [34.5%] vs 322 [64%], p <0.001), and less likely to have received adjuvant chemotherapy (11 [10%] vs 172 [34.2%], p <0.001). All other patient characteristics and breast cancer treatments were comparable between those who initiated ET and those who did not initiate ET.

Table 2: Baseline characteristics of the cohort with hormone receptor-positive breast cancer.

Cohort grouped by endocrine therapy (ET) initiation status.

	Total (n=613)	ET (n=503)	No ET (n=110)	p-value
Age, mean (SD)	58 (12)	57 (11)	59 (14)	0.40
BMI, mean (SD)	29 (6)	29 (6)	28 (6)	0.011
Menopausal status				0.23
Postmenopausal	404 (65.9%)	330 (65.6%)	74 (67.3%)
Premenopausal	195 (31.8%)	167 (33.2%)	28 (25.5%)
Missing	14 (2.3%)	6 (1.2%)	8 (7.2%)
Race				0.95
White	545 (88.9%)	448 (89.1%)	97 (88.2%)
Black	26 (4.2%)	21 (4.2%)	5 (4.5%)
Asian	12 (2.0%)	10 (2.0%)	2 (1.8%)
Other	29 (4.7%)	23 (4.6%)	6 (5.5%)
Missing	1 (0.16%)	1 (0.20%)	0 (0%)
Node Stage				0.023
pN0 or not assessed	448 (73.1%)	357 (71.0%)	91 (82.7%)
pN1-N3	164 (26.8%)	145 (28.8%)	19 (17.3%)
Unknown	1 (0.16%)	1 (0.20%)	0 (0%)
ER status				>0.99
Negative	5 (0.82%)	4 (0.80%)	1 (0.91%)
Positive	606 (98.9%)	498 (99.0%)	108 (98.2%)
Missing	2 (0.33%)	1 (0.20%)	1 (0.91%)
PR Status				0.41
Negative	75 (12.2%)	59 (11.7%)	16 (14.5%)
Positive	534 (87.1%)	441 (87.7%)	93 (84.5%)
Missing	4 (0.65%)	3 (0.60%)	1 (0.91%)
Adjuvant Chemotherapy	183 (29.9%)	172 (34.2%)	11 (10%)	<0.001
Adjuvant Radiation	360 (58.7%)	322 (64.0%)	38 (34.5%)	<0.001
ET (%)
Aromatase inhibitor	332 (54.2%)	332 (66.0%)	-	-
Tamoxifen	153 (25.0%)	153 (30.4%)	-
None/Missing	128 (20.9%)	0 (0%)	-
Fibromyalgia score (SD)	4 (4)	4 (4)	6 (5)	0.027
BPI Worst pain (SD)	1 (2)	1 (2)	2 (2)	0.10
Life satisfaction (SD)	7 (2)	7 (3)	7 (2)	0.47

BMI: body mass index; BPI: Brief Pain Inventory; SD: standard deviation. P-value is comparing the ET and no ET cohorts.

In the ET cohort, the average Fibromyalgia Score prior to surgery was 4/31 (SD 4), with 20 (5.2%) patients having FS ≥ 13. In addition, the average worst pain score was 1/10 (SD 2) and life satisfaction score was 7/10 (SD 3). For those who did not initiate ET, baseline Fibromyalgia Score was 6/31 (SD 5, p=0.027), with 7 (8.0%) patients having FS ≥ 13 (Figure 1). There were no statistically significant differences for other baseline patient-reported outcomes between the two groups. For the 503 patients who initiated ET, we also examined initial ET discontinuation by baseline Fibromyalgia Survey score divided into tertiles. There was a numerically lower likelihood of remaining on the initially prescribed ET for patients in the highest tertile of Fibromyalgia Survey score, although this difference was not statistically significant (p=0.085) (Figure 2).

Figure 1. Distribution of baseline Fibromyalgia Survey [FS] scores amongst hormone receptor-positive patients who initiated anastrozole, exemestane, letrozole, or tamoxifen.

FS score is on the x-axis and total number of patients per each FS score is on the y- axis. Vertical line indicates the FS score cut off for fibromyalgia (FS = 13), with scores to the right of the line consistent with fibromyalgia and scores to the left not meeting criteria for fibromyalgia.

Figure 2. Kaplan Meier curve for likelihood of remaining on initial ET medication, by Fibromyalgia Survey [FS] score tertile.

X-axis indicates the number of patients on endocrine therapy (ET) over the course of five years, for patients in the top-, middle-, and bottom-tertiles of Fibromyalgia Score. Y-axis indicates the likelihood of being on ET.

Univariate analysis of ET discontinuation:

Within the ET group, median duration of follow-up was 1.8 years. Patients who discontinued initial ET and either switched to a different ET or discontinued ET entirely were censored at that time; all other patients were followed until date of data abstraction or completion of ET. Of the 503 who initiated ET, first ET agent was tamoxifen, anastrozole, exemestane, or letrozole for 485 patients. The other18 patients were prescribed toremifene, fulvestrant, or GnRH agonist therapy only (n=14) or their primary tumors were not estrogen receptor positive (n=2), so were excluded from the analysis. There were 5 patients without an identified reason for ET discontinuation who were also excluded from the analysis. Of the 480 who began one of the four specified ET agents and had an identifiable reason for discontinuation, 203 (42.3%) discontinued their initial ET before completing 5 years of therapy. Of the 203 who discontinued, 179 (88.2%) discontinued initial ET because of medication side effects, including 31/52 (59.6%) tamoxifen-treated patients and 148/151 (98%) AI-treated patients. There were 4 patients who discontinued due to resumption of menstrual cycles and/or elevated estradiol. The rest discontinued due to patient preference, disease progression, or breast cancer recurrence.

On univariate survival regression analysis, tamoxifen use and premenopausal status were associated with a decreased hazard of stopping ET (hazard ratio [HR] 0.70, 95% confidence interval [CI] 0.51–0.95, p=0.021 and HR 0.73, 95%CI 0.54–0.99, p=0.040, respectively) (Table 3). Fibromyalgia Score was associated with an increased risk of stopping ET, with each one-point increase in Fibromyalgia Score increasing the hazard ratio by 1.04 (95%CI 1.00–1.08, p=0.043). No other variables were statistically significantly associated with ET discontinuation.

Table 3:

Univariable analysis of variables associated with endocrine therapy (ET) discontinuation.

	Hazard Ratio (95% CI)	p-value
Age	1.01 (1.00, 1.02)	0.17
BMI	1.00 (0.98, 1.02)	0.95
Race
White	reference	-
Other	0.82 (0.50, 1.32)	0.41
ET drug
Aromatase Inhibitor	reference	-
Tamoxifen	0.70 (0.51, 0.95)	0.021
Menopausal status
Postmenopausal	reference	-
Premenopausal	0.73 (0.54, 0.99)	0.040
Chemotherapy	1.08 (0.81, 1.44)	0.59
Radiation	1.21 (0.90, 1.63)	0.21
Fibromyalgia score	1.04 (1.00, 1.08)	0.043
BPI Worst pain	1.03 (0.95, 1.12)	0.47
Life satisfaction	0.99 (0.93, 1.05)	0.65

BMI: body mass index; BPI: Brief Pain Inventory; CI: confidence interval.

Multivariable analysis of ET discontinuation:

Using a multivariable cox proportional hazard model, only Fibromyalgia Score maintained its statistically significant association with increased likelihood of ET discontinuation (HR 1.04, 95%CI 1.01–1.08 p=0.026). This association remained when menopausal status was removed from the multivariable model (HR 1.04, 95%CI 1.00–1.08 p=0.046) (Online Resource 2).

Discussion

Although adjuvant ET improves breast cancer outcomes, up to half of patients stop therapy early, primarily because of side effects. In this retrospective study, we demonstrated that increased baseline nociplastic pain, as measured by Fibromyalgia Survey Score, was associated with premature discontinuation of ET. This finding suggests that central nervous system dysfunction, rather than peripheral inflammation, may underlie development of some ET-emergent toxicity, which could have implications for management approaches.

In our study, a higher proportion of patients who discontinued ET due to side effects were receiving AI therapy. AI therapy is increasingly used over tamoxifen for adjuvant treatment of hormone receptor-positive breast cancer due to its superior efficacy in reducing breast cancer recurrence and mortality [22]. Despite its demonstrated benefit, many patients experience AIMSS, and up to 30% of patients discontinue their AI due to these side effects [23]. To date, AIMSS has most commonly been attributed to nociceptive pain mechanisms. Some have hypothesized that decreased estrogen and dysfunctional inflammation contribute to microscopic bone and cartilage changes [5, 22]. However, consistent laboratory or radiographic evidence supporting either mechanism has not been found, and AIMSS incompletely responds to treatment with non-steroidal anti-inflammatory medications and opioids [5]. We demonstrated an association between baseline nociplastic pain and ET discontinuation, potentially suggesting a centralized pain mechanism underlying the pathogenesis of AIMSS. Patients who experience AIMSS might benefit from treatment modalities used for nociplastic pain, such as education, cognitive behavioral therapy, acupuncture, and serotonin and norepinephrine reuptake inhibitor (SNRI) therapy [24]. Importantly, both the SNRI duloxetine and acupuncture have been found in randomized clinical trials to be beneficial for improving AIMSS [23, 25, 26]. Efficacy of other nociplastic pain treatments for reducing AIMSS have not been reported.

As our understanding of pain evolves, new pain assessment instruments have been created to capture, categorize, and quantify patients’ pain [16, 27, 28]. The BPI, which is one of the most commonly used questionnaires, focuses on overall pain intensity and interference of pain on daily activities [21, 29]. However, BPI has the limitation that the perception of pain intensity varies between patients [30]. The instrument also fails to capture the character of pain (e.g., burning pain or hyperalgesia), timing of pain (episodic versus chronic), or information about locations of pain. These limitations are relevant when characterizing symptoms in patients with widespread, difficult to localize, and centralized pain, such as those with nociplastic pain conditions. The Fibromyalgia Survey was proposed by the American College of Rheumatology in 2010 as a way to address this issue [13]. The instrument has since been validated for evaluation of fibromyalgia, temporomandibular disorders, rheumatoid arthritis, and chronic lower back pain [31–33]. Studies have demonstrated an association between the Fibromyalgia Survey Score, pain, and disability, even when patients have a Fibromyalgia Score below the threshold for official fibromyalgia diagnosis [31, 32]. Our study found that in patients with breast cancer, an increased Fibromyalgia Score prior to surgery is associated with ET discontinuation, with most patients discontinuing their ET due to side effects. These findings highlight the potential of the Fibromyalgia Survey as a tool to evaluate patients’ underlying risk of ET discontinuation. It is also interesting that patients with hormone receptor-positive disease who did not initiate ET had a higher baseline fibromyalgia score compared to those who initiated ET, although our retrospective study was not designed to examine the clinical implications of this finding.

There were several strengths of this study. The first is that patients were identified from a large, ongoing institutional study in which patient reported outcomes were routinely captured immediately prior to surgery from consecutive, unselected patients, and therefore this cohort is more representative of a real-world population. Second, by limiting our study population to those who did not receive neoadjuvant chemo- or endocrine therapy, we examined a more homogenous population and did not have to account for effects of prior treatment on questionnaire responses. Third, we used validated questionnaires, making our findings more reliable and reproducible. However, patients were recruited from a single cancer center, so results and conclusions might differ in a larger, more diverse population. Also, since the study is retrospective, our conclusions are limited to suggesting association, not causation. Finally, in the MGI project, different questionnaires have been collected over time, resulting in missing data. To minimize errors resulting from missing data, we only included questionnaires completed by at least 75% of patients in our models, although some biases may remain.

In summary, nociplastic pain prior to initiation of any local or systemic breast cancer therapy was associated with premature ET discontinuation. This association suggests a mechanism and potential treatment approaches for ET toxicity, particularly musculoskeletal side effects. In addition, it highlights the Fibromyalgia Survey as a tool to identify patients at increased risk of ET discontinuation. Patients with higher baseline Fibromyalgia Survey scores may benefit from adjustments to the standard ET regimen or proactive side effect management to optimize adjuvant ET adherence and persistence. Further study is necessary to confirm the clinical utility of the Fibromyalgia Survey in a larger, more diverse patient population and to evaluate the impact of other nociplastic pain treatments in patients taking ET.

Abbreviations:

AI

aromatase inhibitor

AIMSS

aromatase inhibitor-induced musculoskeletal symptoms

BMI

body mass index

BPI

Brief Pain Inventory

EHR

electronic health record

ET

endocrine therapy

FS

Fibromyalgia Survey

HRQOL

health-related quality of life

IASP

International Association for the Study of Pain

MGI

Michigan Genomics Initiative

SSI

Symptom Severity Index

Data Availability:

The datasets analyzed during the current study are available from the corresponding author on reasonable request.