Introduction

For the majority of adults in the U.S., work is an essential part of life. Breast cancer survivors (BCS) are no exception; an estimated 60–90 % of BCS return to work full-or part-time after completing cancer treatment [1, 2], and a growing number of BCS work throughout their cancer treatment as well [3, 4]. Beyond the simple economic necessity of working, for many survivors, work offers a vital source of self-esteem, structure, and social support [5]. Unfortunately, for many BCS, side effects related to their cancer treatments can interfere with their ability to return to work and regain pre-treatment levels of occupational function [6–8].

One side effect of cancer treatment that has received comparatively little attention in the discussion of work among BCS is chemotherapy-induced peripheral neuropathy (CIPN), a form of nerve damage associated with a number of commonly used cancer therapies. Current estimates suggest that 30 % or more of BCS receiving chemotherapy develop some degree of CIPN during treatment depending on the agents, regimen, and individual risk factors [9, 10]. While research has begun to illustrate the potentially serious impact that CIPN symptoms (CIPN-sx) can have on women’s quality of life [11–13], ability to complete treatment [14, 15], and ability to perform daily activities [16, 17], the impact of CIPN-sx on survivors’ ability to work is only beginning to be investigated.

This is troubling for several reasons. First, women make up nearly half of the U.S. labor force [18] and more than 99 % of cases of breast cancer in the U.S. [19]. In many of these cases, use of neurotoxic chemotherapy is standard, putting more and more BCS at risk for developing CIPN. Second, studies have shown that the majority of BCS resume work after treatment [20] but often with significant difficulty [21]. Issues ranging from fatigue, physical limitations, and cognitive symptoms have all been implicated in this difficulty returning to pre-treatment levels of occupational functioning [5, 20], but the role that CIPN-sx may play in this difficulty is unclear. This lack of clarity is particularly troubling because 23–86 % of BCS who develop CIPN-sx during treatment continue to report CIPN-sx after treatment [22–27], when return to work is likely. Thirdly, while a number of studies have shown that the severity of CIPN-sx is associated with disruptions in treatment [14, 15], poorer quality of life [28], and greater use of health care resources [29], it is not clear whether the presence, frequency, severity, or total number of CIPN-sx is the best predictor of difficulty working post-treatment.

The purpose of this analysis was to evaluate the impact that the presence, frequency, number, and severity of CIPN-sx have on BCS’ perceived ability to work post-treatment. Specific aims were to (1) compare the presence, frequency, number, and severity of CIPN-sx in BCS exposed to chemotherapy (Ctx+) to BCS whose treatment did not include chemotherapy (Ctx−); (2) compare perceived ability to work between Ctx+ and Ctx− and determine whether the presence, severity, and total number of CIPN-sx survivors reported were associated with their ability to work during the first year post-treatment; and (3) explore which combinations of CIPN-sx best predict survivors’ ability to work post-treatment.

Methods

Sample and eligibility

Data for the analysis came from a recent longitudinal study evaluating the effect of cancer and cancer treatment on cognitive function in Ctx+ and Ctx− women with non-metastatic breast cancer (i.e., stages 0–IIIc). The original study also included demographically matched healthy controls, as previously described [30–32]. Healthy controls were not evaluated for CIPN-sx and, therefore, were not included in the analysis.

Participants were recruited from the Indiana University Melvin and Bren Simon Cancer Center’s recruitment core and affiliated clinical sites. Approval for the study was granted by the Indiana University Institutional Review Board. Written informed consent was collected from all participants. Data was collected at the following three time points: (1) baseline (after breast surgery but before radiation, chemotherapy, or anti-estrogen treatment), (2) approximately 1 month (1 M) after completing chemotherapy, and (3) approximately 1 year (1 Y) after the 1 M visit. Approximately one third of Ctx+ received neo-adjuvant chemotherapy and were surgery and treatment naïve at baseline.

The sample for this analysis consisted of 22 Ctx− and 22 Ctx+ women with non-metastatic breast cancer, ages 69 or younger. Exclusion criteria included a self-reported history of prior cancer; substance abuse; and other medical, neurological, and psychiatric risk factors with the potential to affect central or peripheral neurological structure/function [30, 31]. All Ctx+ women were treated with standard doses of chemotherapy agents known to cause CIPN-sx, such as taxanes and platinum compounds.

Measures

Demographics and cancer treatment

Age at baseline (years), education (years), race and ethnicity (categories), and initial stage of breast cancer (0–IIIc) were collected by self-report at baseline. Information on participants’ exposure to chemotherapy and other treatments associated with the development of symptoms simlar to CIPN-sx (e.g., muscle/joint pain) were collected from medical records.

Presence, frequency, number, and severity of CIPN-sx

The presence, frequency, number, and severity of CIPN-sx were measured using the neurotoxicity subscale of the Functional Assessment of Cancer Treatment Gynecological Oncology Group–Neurotoxicity questionnaire (FACT/GOG-Ntx) scale, version 4. The Ntx is an 11-item subscale of the FACT/GOG that measures the presence and severity of several common CIPN-sx. Items ask participants to rate the degree of sensory, motor, auditory, and functional CIPN-sx they experienced during the past week on a five-point Likert scale (0–4), corresponding with increasing symptom severity [33, 34]. The reliability and sensitivity of the Ntx subscale for assessing CIPN-sx has been established (Cronbach’s α = 0.64–0.86) [35], especially in patients receiving taxanes [34]. Three Ctx+ and four Ctx− included in the original study were excluded from the analysis because complete data on CIPN-sx or perceived ability to work was not available. Given the unique nature of each question, missing data was not imputed.

The presence of CIPN-sx was determined by calculating the total number of Ctx+ and Ctx− women who reported having a specific CIPN-sx, regardless of its severity (i.e., any score greater than 0). The frequency of CIPN-sx was determined by converting the total number of women who reported having each CIPN-sx at each time point into a percentage. The total number of CIPN-sx at each time point was calculated by tabulating the number of unique CIPN-sx participants reported at each time point, regardless of their severity. The severity of CIPN-sx was determined by calculating mean scores for the (a) total FACT/GOG-Ntx scale; (b) total scores for the sensory, motor, hearing, and functional domains; and (c) scores for individual symptoms in each domain at each time point.

Perceived ability to work and employment status

Participants’ perceived ability to work was measured using an item from the Functional Well-Being subscale of the FACT/GOG-Ntx (version 4), which asked participants to respond to the statement “I am able to work (including housework).” Perceived ability to work was scored on a five-point Likert scale (0 = not at all, 1 = a little bit, 2 = somewhat, 3 = quite a bit, 4 = very much).

To help inform the discussion about the impact of CIPN-sx on occupational function, we also evaluated (1) the percentage of women that were working either full-or part-time at each time point, (2) the type of work in which they were engaged (unskilled, semi-skilled, skilled, managerial/clerical/official/sales, or professional/technical), and (3) any change in the type of work performed from baseline to 1 Y.

Statistical analyses

Analyses were performed using SPSS, version 23 (IBM Corporation). Descriptive statistics were used to tabulate the presence (yes/no) and frequency (%) of CIPN-sx at each time point. Pearson’s chi-squared tests were used to compare the frequency of categorical demographic variables and the frequency of CIPN-sx between groups at 1 M and 1 Y. For categorical variables with fewer than five observations, Fisher’s exact test was substituted. For categorical variables with fewer than five observations where order mattered (i.e., stage of cancer and perceived ability to work), Mantel-Hanzel chi-squared tests were used. Independent sample t-tests were used to compare continuous demographic variables, the number of CIPN-sx reported, and total FACT/GOG-Ntx scores. Differences in domain scores and individual items on the FACT/GOG-Ntx were compared using Mann-Whitney U tests, with a Benjamini-Hochberg correction to reduce the risk of type I errors due to multiple comparisons. Differences in ordinal responses to the work item between Ctx+ and Ctx− were compared using ordinal regression models, using cancer stage as a covariate. Spearman’s coefficients (R s ) were used to identify significant correlations between the presence, severity, and number of CIPN-sx and ability to work post-treatment.

Ordinal regression models were constructed to explore the combination of CIPN-sx that best predicted ability to work for Ctx+ at 1 M and 1 Y. Because of the small sample size, 1 M and 1 Y models were limited to two CIPN-sx apiece (i.e., one predictor variable per ∼10 observations). Only the severity of CIPN-sx was used in the analyses to minimize the potential to inflate significance levels because of variables with shared sources of variance. Final regression models were selected based on their statistical significance (p value) and explanatory value (pseudo r 2 value). Because of the sample size, models were also carefully screened for signs of overfitting (e.g., initial model fit <0.05, goodness of fit >0.05).

Results

Demographics

Table 1 summarizes the demographics for the sample. Participants were predominantly middle-aged, white, and well-educated. Groups did not differ significantly on age, race, education, use of radiation, amount of radiation received, use of hormonal or biologic therapies, or exposure to other agents commonly given during breast cancer treatment associated with painful or neuropathy-like symptoms (Table 1). Groups differed only on stage of cancer and use of trastuzumab.

Table 1 Sample characteristics
Full size table

Table 1 also details participant’s exposure to cancer treatments associated with CIPN-sx or similar symptoms (e.g., muscle/joint pain). More than 95 % of Ctx+ received at least one neurotoxic agent during treatment, and almost a third received two neurotoxic agents. Of these, all but one received a taxane (docetaxel 54.5 %, paclitaxel 40.9 %). Mean taxane exposure was 911 ± 492 mg/m2. Five Ctx+ who received a taxane also received the platinum compound carboplatin. Mean platinum exposure among Ctx+ was 958.1 ± 1828.9 mg/m2.

Presence and frequency of CIPN-sx

At baseline, Ctx+ and Ctx− did not differ significantly on the presence, frequency, severity, or total number of CIPN-sx (Tables 2 and 3). At 1 M, more than 50 % of Ctx+ reported numbness, tingling or discomfort in their hands or feet; joint pain/muscle cramps; weakness; and difficulty feeling the shape of small objects. With the exception of three symptoms (joint pain/muscle cramps, trouble walking, and trouble hearing), Ctx+ reported CIPN more frequently than Ctx− for all symptoms we evaluated at 1 M. At 1 Y, the frequency and type of CIPN-sx reported by Ctx+ was very similar to 1 M, with more than half of Ctx+ reporting numbness, tingling, or discomfort in their hands or feet; joint pain/muscle cramps; and weakness.

Table 2 Frequency and total number of chemotherapy-induced peripheral neuropathy symptoms (CIPN-sx) at baseline, 1-month, and 1-year time points
Full size table
Table 3 Severity of chemotherapy-induced peripheral neuropathy symptoms (CIPN-sx) at baseline, 1-month, and 1-year time points
Full size table

Between 1 M and 1 Y, the frequency of some CIPN-sx increased slightly between for Ctx+. At 1 Y, a slightly higher percentage of Ctx+ reported hand numbness/tingling (63.6 vs. 59.1 %), foot numbness/tingling (57.1 vs. 54.5 %), joint pain/muscle cramps (72.7 vs. 68.2 %), trouble buttoning buttons (33.8 vs. 31.8 %), and difficulty walking (38.1 vs. 27.3 %) than at 1 M.

Total number of CIPN-sx

At 1 M, Ctx+ reported an average of 5.59 ± 3.5 CIPN-sx, compared to 3.14 ± 2.0 CIPN-sx for Ctx− (p = 0.006; Table 2). A year later, the total number of CIPN-sx Ctx+ reported was virtually unchanged; at 1 Y, Ctx+ reported an average of 5.18 ± 3.5 symptoms and Ctx− reported 3.41 ± 2.3 (p = 0.049).

Severity of CIPN-sx

At 1 M, Ctx+ reported significantly more severe functional CIPN-sx than Ctx− (p = 0.007). In addition, total scores on the sensory domain just missed the cutoff for statistical significance after adjusting for multiple comparisons (p = 0.006; Table 3). Inspection of individual FACT/GOG-Ntx scores revealed that Ctx+ had more severe numbness/tingling in their hands (p = 0.002) and feet (p = 0.001), hand discomfort (p = 0.012), weakness (p = 0.003), ringing/buzzing in their ears (p = 0.009), and trouble buttoning buttons (p = 0.004) than Ctx− after treatment.

At 1 Y, while total and sensory FACT-GOG/Ntx scores were not significantly different between Ctx+ and Ctx−, Ctx+ continued to report more severe functional CIPN-sx than Ctx− (p = 0.004), including more severe numbness/tingling in their hands (p = 0.001) and feet (p = 0.013) and trouble buttoning buttons (p = 0.002; Table 3).

Employment status and perceived ability to work

At baseline, 1 M, and 1 Y, all 44 Ctx+ and Ctx− were working part-or full-time (Fig. 1). Of these, 92.4 % of Ctx+ and more than 70 % of Ctx− were working in positions classified as either professional/technical or managerial/official/clerical/sales in nature (Fig. 2). Over the course of the study (i.e., BL to 1 Y), there was virtually no change in the type of work participants performed (data not shown).

Fig. 1
figure 1

Employment status, type of work, and perceived ability to work at baseline, 1-month, and 1-year time points. The figure illustrates the percentages of Ctx+ and Ctx−, respectively, who were classified as working in each type of work (unskilled, semi-skilled, skilled, managerial/clerical/official, or professional/technical) at each time point (BL, 1 M, 1 Y)

Full size image
Fig. 2
figure 2

Perceived ability to work at baseline, 1-month, and 1-year time points. The figure illustrates the percentage of Ctx+ and Ctx− who reported being not at all, a little bit, somewhat, quite a bit, and very much able to work (including housework) at each time point

Full size image

Compared to survivors who reported being “very” able to work at 1 M, having received chemotherapy during treatment significantly predicted being only “somewhat” (p = 0.00), “a little bit” (p = 0.00), or “not at all” (p = 0.03) able to work, after controlling for differences in cancer stage between Ctx+ and Ctx− (data not shown). Similarly, at 1 Y compared to women who were very able to work, having received chemotherapy during treatment was also significantly associated with being only “a little” or somewhat able to work, after controlling for stage of cancer. This difference was especially evident at 1 M, where 50 % of Ctx+ reported being only somewhat, a little bit, or not at all able to work, compared to just 9.1 % of Ctx− survivors.

Association between presence, severity, and number of CIPN-sx and ability to work

At 1 M, the severity of the following five combinations of CIPN-sx were correlated with BCS’ work ability: hand numbness/tingling (R s  = −0.483; p = 0.023), hand discomfort (R s  = −0.511; p = 0.015), weakness (R s  = −0.557; p = 0.007), trouble hearing (R s  = −0.454; p = 0.034), and difficulty feeling the shape of small objects (R s  = −0.463; p = 0.030). In addition, at 1 M the presence of any hand discomfort (R s  = −0.455; p = 0.033), weakness (R s  = −0.603; p = 0.003), trouble hearing (R s  = −0.501; p = 0.018), or trouble feeling the shape of small objects in hand (R s  = −0.433; p = 0.044), regardless of severity, were also significantly correlated with work scores among Ctx+, as was the number of CIPN-sx reported at 1 M (R s  = −0.526; p = 0.012).

At 1 Y, only the presence of weakness (regardless of severity) was associated with perceived work ability for Ctx+ (R s  = −0.478; p = 0.024).

Using the severity of CIPN-sx to predict work ability post-treatment

Results of the exploratory analysis using ordinal regression identified five models that predicted Ctx+ who were not at all able to work at 1 M, which are presented in Table 4: (1) hand numbness and trouble feeling the shape of small objects (Wald χ 2(1) = 11.39; cumulative OR = 0.008; Nagelkerke r 2 = 0.500), (2) trouble buttoning buttons and trouble feeling the shape of small objects (Wald χ 2(1) = 12.99; OR = 0.004; Nagelkerke r 2 = 0.567), (3) foot numbness and foot pain (Wald χ 2(1) = 7.65; OR = 0.031; Nagelkerke r 2 = 0.644), (4) foot numbness and trouble walking (Wald χ 2(1) = 10.67; OR = 0.003; Nagelkerke r 2 = 0.724), and (5) trouble hearing and hand pain (Wald χ 2(1) = 12.02; OR = 0.005; Nagelkerke r 2 = 0.583). At 1 Y, no combination of CIPN-sx significantly predicted work ability for Ctx+ (data not shown).

Table 4 Ordinal regression models using the severity of CIPN-sx to predict perceived work ability in chemotherapy-treated (Ctx+) breast cancer survivors approximately 1 month after treatment (N = 22)
Full size table

Discussion

There is growing evidence that CIPN-sx pose a risk to BCS not only during cancer treatment but after as well [24–26, 36–41]. Previous research has linked CIPN-sx with poorer quality of life and decreased ability to function [28], but this is one of the first studies to address the question of whether CIPN-sx impact work in BCS post-treatment [28]. Results of this study add to the growing evidence showing the persistence of CIPN-sx after treatment [25, 26, 36], and extend these findings to the context of occupational function, demonstrating a clear link between the presence, frequency, number, and severity of CIPN-sx and BCS’ self-reported inability to work following treatment. Results of our analysis also suggest that (1) the total number of CIPN-sx Ctx+ report, regardless of their severity, may be a useful predictor of difficulty working and that (2) painful and non-painful CIPN-sx affecting the hands or feet (with or without treatment-related hearing deficits) may predict work difficulty for Ctx+ 1 M post-treatment.

Presence and frequency of CIPN-sx

An important finding from our analysis was that the pattern of CIPN-sx reported by Ctx+ approximately a year post-treatment was very similar to the pattern reported at 1 M. In the context of work, this has important implications because it raises the possibility that the CIPN-sx survivors present with immediately after treatment may be a reliable indicator of the CIPN-sx that will continue to interfere with a survivor’s ability to work during the critical first year of survivorship. Research in larger samples is needed to test this hypothesis.

Total number of CIPN-sx

A second potentially important finding from our analysis was that the number of CIPN-sx Ctx+ reported, independent of their severity, appeared to predict ability to work post-treatment. The majority of studies have used either the presence [14, 15, 29] or severity of CIPN-sx as the primary lens for evaluating CIPN-related outcomes [41]. While our results suggest that both are likely to predict work ability, the number of CIPN-sx patient’s experience itself may also be disruptive to work. On average, Ctx+ in our sample reported five different CIPN-sx at both post-treatment time points. As with other studies, when we looked at the severity of these CIPN-sx, in many cases, we observed a pattern of one or two severe CIPN-sx and several milder symptoms [41]. Because this mixed severity pattern is common, it will be important to clarify whether the number of CIPN-sx survivors report is just as predictive of difficulty working as severity of each symptom. This is not only because the severity of CIPN-sx can vary but also because CIPN-sx often grow less severe over time [10], which could lead providers to overlook the potentially serious impact that a number of milder CIPN-sx could have on BCS’ ability to work.

Association between the severity of CIPN-sx and ability to work at 1 M

During our regression analyses, we identified several combinations of CIPN-sx that were predictive of Ctx+ that were less able to work at 1 M. These included two combinations of sensory CIPN-sx affecting the hands, two combinations of CIPN-sx affecting the feet, and the combination of hearing loss and hand discomfort (Table 4). These findings are consistent with the few studies of CIPN in which work was considered [28] and make sense intuitively. In particular, the finding that sensory symptoms that interfere with women’s ability to feel the shape of objects or button buttons were associated with difficulty working was not surprising. These combinations of CIPN-sx point to potential phenotypes for CIPN-related work interference. It is important to note that several other symptoms such as weakness and joint pain/muscle soreness also predicted inability to work at 1 M in several of our models but were not included because of lack of model fit given our sample size. Statistical considerations notwithstanding, it is clear that symptoms such as joint pain/muscle soreness and weakness have the potential to impact work and should be included in future studies of CIPN-sx and work.

Association between the severity of CIPN-sx and ability to work at 1 Y

There are several reasons that may explain why we did not observe a stronger effect of CIPN-sx on perceived work ability at 1 Y. First, the percentage of Ctx+ who reported being “very much” or “quite a bit” able to work after treatment rose from 41.4 % at 1 M to 90.9 % at 1 Y (Fig. 2a). At the same time, the type, frequency, severity, and number of CIPN-sx Ctx+ reported remained relatively constant. This combination of improving ability to work in the face of relatively stable CIPN-sx suggests that while BCS may continue to experience CIPN post-treatment, the impact of these symptoms on work may lessen over time as BCS acclimate to their symptoms and/or develop coping strategies.

Finally, it is important to note that while the lack of treatments for CIPN-sx is problematic on many fronts, it is particularly concerning in the context of work. Recently, several groups have had some success treating the painful component of CIPN-sx non-invasively using devices such as scrambler therapy [42], as well as several oral or topical agents [10]. These successes raise the question of whether strategies like these could be used to address painful CIPN-sx before they can disrupt cancer survivors’ transition back to usual work activities.

Limitations

While this study provides a useful starting point for future research exploring the impact of CIPN-sx on breast cancer survivor’s ability to work post-treatment, our results need to be considered in light of several limitations. First, the sample for this analysis was small; larger samples will be needed to validate these findings and explore which combinations of CIPN-sx best predict survivor’s ability to work. Secondly, we did not have access to detailed information on potential risk factors for CIPN such as osteoarthritis, and the size of our sample prevented us from including symptoms such as depression as covariates in our analysis, both of which should be included in future studies. Third, our analysis relied on a single item to evaluate perceived ability to work post-treatment. While this item provided a useful lens for looking at the impact of CIPN-sx on work, clearly, more stringent and varied measures of work including absenteeism, productivity, and performance on job-specific tasks will be needed to understand the occupational impact of CIPN-sx. In addition, we did not have access to objective measures of CIPN, which would have provided valuable insight into structural or functional changes in nerves that may help to explain perceived difficulties working. Fourth, our sample was racially/ethnically homogeneous; more diverse cohorts will be needed to understand whether the impact of CIPN-sx on work outcomes differs by race, ethnicity, or culture. Finally, our sample was approximately 10 years younger than the median age of BCS in the USA. The reason for this was that the study upon which our analysis was based enrolled only women younger than age 70. As such, while our sample is representative of working-age BCS, studies in older survivors that continue work past the current typical age of retirement will be needed to explore the impact of CIPN-sx on work in this population fully.

Conclusion

The increasing reliance on neurotoxic chemotherapy to treat many forms of breast cancer means that clinicians and patients will need to carefully consider how best to balance the benefits and risks of treatment, including the potential impact of CIPN-sx on their ability to work. Our findings suggest that clinicians should remain vigilant for CIPN-sx that may interfere with women’s ability to work during the first year post-treatment, when many survivors return to work. Women who continue to experience CIPN-sx post-treatment (especially in the context of other symptoms that can interfere with work such as pain, fatigue, or cognitive disruption) may need to be referred to physical or occupational therapy to minimize the negative impact of CIPN-sx on work.

This concern may be especially pertinent in the modern workplace, with its growing reliance on tactile technologies such as keyboards and touch screens, which require users to be able to effortlessly tap, touch, type, and glide their way through their workday. Identifying the specific CIPN-sx and dimension of the symptom experience (i.e., presence, frequency, severity, or total number) that best predict difficulty working is an essential first step towards developing interventions to reduce their impact on work for BCS with CIPN.