1. Introduction
Paclitaxel is among the most commonly used cytostatic drugs in breast cancer treatment. In combination with anthracyclines, paclitaxel significantly reduces breast cancer mortality [
1]. Chemotherapy-induced peripheral polyneuropathy (CIPN) is one of the most common dose-limiting side effects of taxanes and can significantly impair quality of life (QoL) and may lead to less effective anticancer treatment [
2,
3,
4]. A meta-analysis of 31 studies with 4179 patients showed a prevalence of CIPN of 68.1% measured 3 months after chemotherapy [
5]. CIPN can affect the peripheral nervous system from the nerve cell to the axon and all types of nerve fibers. However, paclitaxel mainly causes sensory axonal damage, characterized by numbness, tingling, abnormal temperature sensations, or a burning sensation in the fingers and toes [
6,
7]. Currently, there is no approved therapy for CIPN and there is a lack of pharmacological prevention strategies. In practice, local cryotherapy (cooling gloves and socks) or local compression are the most common physical methods of prevention, yet the efficacy is conflicting [
8]. Detection is based on subjective criteria, which indicate CIPN only when symptoms are present [
9]. Therefore, early detection methods are necessary to allow for therapy adjustment to prevent permanent damage. However, current methods of assessing small fibers are invasive, time-consuming, or subjective.
The cornea is one of the most densely innervated tissues, and corneal sensory nerves are considered part of the peripheral nervous system [
10]. Thus, the cornea could represent a diagnostic window for neuropathic changes [
11,
12,
13]. The sensory innervation of the cornea occurs mainly via the trigeminal ganglion in the form of Aδ and C fibers responsible for nociceptive, chemical, and temperature signaling. Although the innervation of the cornea is similar to the receptors in the skin of the distal extremities of the limb [
14,
15], limited research on corneal nerve fiber changes associated with CIPN exists. Recent studies have shown that confocal microscopy can visualize small fibers of the cornea in vivo and detect peripheral neuropathy [
16]. Corneal confocal laser scanning microscopy (CLSM) is an established method for detecting and monitoring diabetic neuropathy [
17]. However, there is low evidence for corneal nerve changes associated with cancer therapy and CIPN. In a recent study of 21 patients with upper gastrointestinal cancer, corneal confocal microscopy could detect a small fiber neuropathy which was related to the severity of CIPN [
18]. Moreover, Chiang et al. [
19] reported reduced corneal nerve fiber and inferior whorl length in paclitaxel-treated patients compared to healthy controls. Riva et al. [
20] tested if the combination of corneal nerve parameters assessed by in vivo confocal microscopy (IVCM) is useful to monitor the neurotoxic effects of chemotherapy compared to epidermal nerve quantification. For this purpose, 95 adults with different cancer types were recruited, and neurological examinations, including a total neuropathy score and
IVCM, were performed before and after chemotherapy [
20]. The experimental data confirmed that in vivo confocal microscopy is a helpful, non-invasive tool with promise for the diagnosis of CIPN [
20]. However, at the present time, despite a case report [
21], no prospective longitudinal studies which monitor breast cancer patients before, during, and after the course of chemotherapy exist for detecting corneal nerves by using advanced biophotonic techniques. In addition, currently, representative selections of CLSM images and valid image analyses are a challenge.
Apart from corneal sensory nerves, the posterior segments of the eye, especially the retina and the choroid, also represent important targets for the toxicity of anticancer drugs [
22]. Anatomically and developmentally, the retina is an extension of the central nervous system and is composed of layers of specialized interconnected neurons [
23]. The retina is among the most metabolically active tissues in the body, making it a prime target for the side effects of chemotherapeutic agents, e.g., macula edema, photopsia, and glaucoma [
22,
24,
25]. In many neurological disorders, ocular manifestations precede neurological symptoms, and thus, retinal examination offers a unique opportunity for the early diagnosis of the side effects of neurotoxic drugs [
22]. In this context, optical coherence tomography (OCT) provides excellent non-invasive access to the retina.
In order to help guide the appropriate treatment plan for each patient and to have a significant influence on their outcome and QoL, there is a clinical need for the early detection of neurotoxic side effects. To address this, the present study aims at determining whether neurotoxic signs in breast cancer patients receiving paclitaxel (taxane) correlate with retinal or corneal nerve changes. To this end, non-invasive in vivo biophotonic imaging methods such as corneal CLSM and OCT will be applied before, during, and after the cancer therapy regimen.
2. Materials and Methods
2.1. Study Design and Flow Chart
The prospective longitudinal study was approved by the Ethics Committee of the University of Rostock in accordance with applicable laws, rules, and regulations on 19 February 2020 (A 2018-0162). All methods were conducted in accordance with relevant guidelines and regulations. All participants underwent informed consent and signed a consent form prior to the conduction of the study in accordance with the Declaration of Helsinki. All patients underwent monitoring sessions after diagnosis and before (T0), during (T1 and T2), and after cessation (T3) of the paclitaxel therapy regimen. The monitoring sessions included the assessment of QoL and neurological scores, ophthalmological status, and anterior and posterior segment imaging (
Figure 1). Fourteen patients with newly diagnosed breast cancer and ten healthy controls were recruited. Of the 14 recruited patients, 2 were excluded due to health reasons, therefore, 12 patients and 10 control subjects were assessed. The 12 patients were able to participate in the clinical examinations depending on their therapy. Because of different therapy durations and individual health conditions, not all patients could be examined at all ophthalmological monitoring sessions.
Figure 1 summarizes the total number of patients and controls who underwent the monitoring sessions from T0 to T3.
2.2. Patient Selection
Patients were eligible if the following criteria were fulfilled: planned paclitaxel-based chemotherapy for ≥12 weeks, aged 18+ years, and willing to complete all study activities. Patients were excluded if <1/3 of planned neurotoxic chemotherapy was applied, or if they had a pre-existing neuropathy due to other reasons (e.g., diabetes mellitus or alcohol consumption). Healthy controls had no history of cancer, no self-reported signs of peripheral neuropathy, and no ocular diseases.
2.3. Assessment of Clinical Characteristics
Patients’ and controls’ age, BMI, history of the disease (ophthalmological, neurological, gynecological, oncological, and diabetes mellitus), and additional clinical data (e.g., allergies, current medication, and nicotine, drug, and alcohol consumption) were obtained by interview or chart review. Additionally, patients’ tumor-specific data, such as the initial stage, grading and hormone receptor status, and cumulative dose of chemotherapeutic agents were gathered.
2.4. Patient-Reported Outcome Measures and Neurological Scores
Patients completed the CIPN toxicity assessment at different time intervals: at baseline before therapy (T0), after anthracycline/cyclophosphamide therapy (T1), after 6 courses of the paclitaxcel treatment (T2), and after 12 courses (T3) of weekly 90 mg/m
2 paclitaxel (or fewer courses if therapy was aborted early due to side effects) (
Figure 1). All neurological assessments were performed by one previously trained researcher. Clinical data were obtained from the patients’ medical records. For patient-reported outcome (PRO) measures, standardized questionnaires were used:
- -
The European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ)-CIPN20 module, a 20-item self-reporting questionnaire containing three subscales to assess sensory, motor, and autonomic CIPN [
26]. The total score ranges from 20 to 80; depending on sex and car driving behavior the range can decrease from 18 to 72; higher scores indicate worse CIPN.
- -
The Functional Assessment of Cancer Therapy/Gynecologic Oncology Group-Neurotoxicity (FACT/GOG-NTX) with subscales of physical, social, emotional, and functional wellbeing. The individual scores range from 0 to 28.
- -
The Taxane subscale (TaxS) is a 16-item self-reporting questionnaire focusing on patient-reported neurotoxicity symptoms and concerns. The individual items are scored from 0 to 4 and the sum score ranges from 0 to 64; lower scores indicate worse CIPN [
27].
- -
The Trial Outcome Index (TOI) of the FACT-Taxane questionnaire is calculated from physical and functional-wellbeing in addition to the taxane subscale. The score ranges from 0 to 120, with lower scores indicating higher CIPN.
- -
All subscores (physical, social, emotional, functional well-being, and Taxane subscale) together result in the FACT-Taxane Total Score, which ranges from 0 to 172.
Furthermore, neurologic features were evaluated by the Neuropathy Symptom Score (NSS) and Neuropathy Disability Score (NDS) [
28]. The NSS includes questions regarding CIPN typical symptoms and scores from 0 to 10. Scores ranging from 0 to 2 were regarded as normal, 3 to 4 as mild symptoms, 5 to 6 as moderate symptoms, and 7 to 10 as severe neuropathy symptoms. The NDS contains typical neurological qualities that can be limited by CIPN. It scores from 0 to 10; scores ranging from 0 to 2 were regarded as normal, 3 to 5 as mild neuropathic signs, 6 to 8 as moderate neuropathic signs, and 9 to 10 as severe neuropathy.
2.5. Assessment of the General Ophthalmological Status
All 12 patients received complete ophthalmological examinations of both eyes, including the determination of the best-corrected distance visual acuity (BCVA) in decimal and converted to logMAR (Logarithm of the Minimum Angle of Resolution), intraocular pressure (ICare tonometer, Icare Finland Oy, Vantaa, Finland), and slit-lamp examinations including a fundoscopy without mydriasis (BM 900 Haag Streit, Möller-Wedel Optical GmbH, Wedel, Germany). Furthermore, patients underwent corneal densitometry with a Scheimpflug system (Pentacam HD, Oculus Optikgeräte GmbH, Wetzlar, Germany) assessment of the autofluorescence of the lens (ClearPath DS-120
® Biomicroscope, Freedom Meditech, San Diego, CA, USA) [
29,
30] and corneal esthesiometry (Cochet-Bonnet, The Luneau Technology Group, Larché, France). Controls received general ophthalmological examinations only. All examinations were performed by the same experienced ophthalmologist throughout the study.
2.6. Assessment of Posterior Retinal Segment Imaging and Quantitative Retinal Thickness Analysis
Macular imaging (fundus imaging) was performed using a laser-based imaging device combining OCT and scanning laser ophthalmoscopy (SPECTRALIS, Heidelberg Engineering GmbH, Heidelberg, Germany) as described before [
31,
32,
33]. Briefly, the volume scan was sized to cover a rectangular area, centered on the fovea, and included consecutive cross-sectional scans. For each examination, care was taken that the size of the volume covered the area for the commonly used early treatment diabetic retinopathy study (ETDRS) grid [
34]. After retinal layer segmentation and image registration, dedicated deviation maps were generated by applying an in-house developed visual analysis software to the segmented volume data to visualize and quantify locally resolved thickness differences between selected datasets [
32,
35,
36]. A schematic overview of the generation of deviation maps is shown in
Figure 2. The deviation maps were used to compare the retinal thickness of patients at each follow-up time point (T1 to T3) with the baseline time point (T0).
2.7. Assessment of Anterior Corneal Segment Imaging and Quantitative Corneal Nerve Analysis
A combination of the Heidelberg Retina Tomograph 3 (HRT3, Heidelberg Engineering GmbH, Heidelberg, Germany), the Rostock Cornea Module 2.0 (RCM 2.0) [
37,
38], and EyeGuidance system [
39,
40] was used for the in vivo confocal laser scanning microscopy (CLSM) assessment of large areas of the corneal subbasal nerve plexus (SNP). Before the measurement, both eyes were anesthetized with topical anesthetic (Proparakain-POS
® 0.5% eye drops, Ursapharm, Saarbrücken, Germany) and covered with Vidisic Gel (Bausch & Lomb/Dr. Mann Pharma, Berlin, Germany). This gel was used as an artificial tear and an immersion medium.
During the continuous CLSM image acquisition, the EyeGuidance system presents a fixation target moving on an extending spiral path to the contralateral eye, thus leading to a smoothly guided movement of both eyes. Simultaneously, the focal plane is oscillated by the RCM 2.0 centered on the (manually chosen) initial position at the SNP level with ±20 µm amplitude with a constant speed of 120 µm/s. This imaging procedure generates a volumetric dataset with sufficient SNP images for successful large-area mosaicking. In the mosaicking process, every type of corneal image (epithelium, SNP, and stroma) is used for elastic image registration. Then, a tissue classification algorithm excludes every non-SNP image for the final mosaic. This method has been previously described in detail [
41] and is now capable of real-time mosaicking [
13]. A schematic overview of the imaging and mosaicking process is shown in
Figure 3.
The subsequent image processing and quantitative image analysis were performed by a dedicated algorithm using Mathematica (Version 11.3, Wolfram Research Inc., Champaign, IL, USA), as previously described [
42]. The following SNP quantities (
Table 1) were calculated: corneal nerve fiber length (CNFL), corneal total nerve fiber density (CTNFD), corneal nerve branching point density (CNBPD), and corneal nerve fiber tortuosity (CNFTo).
In this longitudinal study, care was taken to evaluate SNP parameters at each time point from the same regions within each patient’s SNP mosaic. For the registration of longitudinal SNP mosaics, static landmarks of the cornea, such as Kobayashi structures [
43,
44] and nerve entry points, were used as previously described in [
45].
2.8. Data Management and Statistical Analysis
Data were analyzed using IBM SPSS 27.0 (Armonk, NY, USA). Significance was defined as p < 0.05. Normality was considered when p ≥ 0.05 in the Kolmogorov–Smirnov test. All continuous data were expressed as the mean ± SD or as the median (range) for nonparametric samples. Independent sample t-tests (Mann–Whitney-U test for nonparametric samples) were used to demonstrate the differences between patients and the control group. Friedman’s test (for nonparametric samples) was used to evaluate the differences in QLQ and neuropathic scores between patients at baseline and during therapy (follow-up). Categorical variables were reported in frequencies. Fisher’s exact test or Pearson’s Chi-square tests were used to evaluate the association between categorical variables.
An Analysis of Variance (ANOVA) with repeated measures was used to analyze the longitudinal treatment effects on morphological SNP parameters.
4. Conclusions
The herein-conducted study represents the first longitudinal study combining the oncological examinations of CIPN symptoms with advanced ophthalmological biophotonic imaging techniques. Thus, taxane-based therapy-induced side effects could be directly compared with each other, representing the first longitudinal study with a constant examination of oncological CIPN parameters based on PROs and NDS/NSS, as well as ophthalmological parameters such as corneal nerves and thickness of retinal layers before, during, and after paclitaxel therapy. The OCT measurements revealed retinal thickenings, representing one key breakthrough of the present study. Moreover, this is the first report detecting increasing retinal thickenings in a longitudinal cancer therapy setting, showing that surveillance protocols for regular eye examinations cannot be overestimated even in asymptomatic cancer patients. The second breakthrough of the study was the longitudinal examination of the SNP nerve parameters in mosaics exceeding the size of standard CLSM images. Furthermore, it was possible to identify and analyze identical regions within the SNP within all monitoring sessions. The herein-applied CLSM technology opens the window for future long-term studies in the setting of neuropathy and its effects on peripheral nerves such as corneal nerves. In summary, regular non-invasive in vivo OCT and CLSM monitoring of a patient’s cornea during cancer therapy could complement routine oncological examinations and be helpful in the generation of a more comprehensive clinical picture with benefits for both oncology and ophthalmology. Regarding the future trend of personalized cancer therapy regimens, biophotonic imaging techniques could represent a powerful diagnostic tool for the objective assessment of the severity of adverse events and the outcome of the patient with retinal and corneal structures acting as potential biomarkers.