Evaluating the cytotoxicity of Ge–Sb–Se chalcogenide glass optical fibres on 3T3 mouse fibroblasts

In vivo cancer detection based on the mid-infrared molecular fingerprint of tissue is promising for the fast diagnosis and treatment of suspected cancer patients. Few materials are mid-infrared transmissive, even fewer, which can be converted into functional, low-loss optical fibres for in vivo non-invasive testing. Chalcogenide-based glass optical fibres are, however, one of the few. These glasses are transmissive in the mid-infrared and are currently under development for use in molecular sensing devices. The cytotoxicity of these materials is however unknown. The cytotoxicity of Ge–Sb–Se chalcogenide optical glass fibres on 3T3 mouse fibroblast cells is here investigated. Fibres exposed to four different pre-treatment conditions are used: as-drawn (AD), propylamine-etched (PE), oxidised-and-washed (OW) and oxidised (Ox). To achieve the latter two conditions, fibres are treated with H2O2(aqueous (aq.)) and dried to produce a surface oxide layer; this is either washed off (OW) or left on the glass surface (Ox). Cellular response is investigated via 3 day elution and 14 day direct contact trials. The concentration of the metalloids (Ge, Sb and Se) in each leachate was measured via inductively coupled plasma mass spectrometry. Cell viability is assessed using the neutral red assay and scanning electron microscopy. The concentration of Ge, Sb and Se ions after a 3 day dissolution was as follows. In AD leachates, Ge: 0.40 mg L−1, Sb: 0.17 mg L−1, and Se: 0.06 mg L−1. In PE leachates, Ge: 0.22 mg L−1, Sb: 0.15 mg L−1, and Se: 0.02 mg L−1. In Ox leachates, Ge: 823.8 mg L−1, Sb: 2586.6 mg L−1, and Se: 3750 mg L−1. Direct contact trials show confluent cell layers on AD, PE and OW fibres after 14 days, while no cells are observed on the Ox surfaces. A >50% cell viability is observed in AD, PE and OW eluates after 3 days, when compared with Ox eluates (<10% cell viability). Toxicity in Ox is attributed to the notable pH change, from neutral pH 7.49 to acidic pH 2.44, that takes place on dissolution of the surface oxide layer in the growth media. We conclude, as-prepared Ge–Sb–Se glasses are cytocompatible and toxicity arises when an oxide layer is forced to develop on the glass surface.


Introduction
Over 100 different types of cancers occur in humans, and only four: breast, colon, prostate, and lung, are responsible for more than half of all diagnoses, with a comparable proportion of deaths. Cancer diagnosis has however improved in recent decades, with advancements occurring in for example, genetic proling and diagnostic imaging. A relatively novel method is mid-infrared hyperspectral imaging. 1 This is based on utilising mid-infrared (MIR) radiation to identify the presence and distribution of chemical species in biological tissue. When a tissue sample absorbs radiation in the MIR region (3.0 mm to 50 mm) 2 covalently bonded bio-molecular species vibrate to give absorption bands whose absorption and characteristic wavelength, provide information pertaining to the quantity, and nature, of their bonding, respectively. 3 The rapidity and high diagnostic accuracy of MIR hyperspectral imaging has been demonstrated by many, for instance: Old et al. 4 and Pilling et al. 5 Old et al. collected and rapidly processed 45 MIR hyperspectral images of oesophageal tissue (5 minutes per image), then accurately identied neoplastic Barrett's oesophagus with 95.6% sensitivity and 86.4% specicity. 4 Similarly, Pilling et al. collected MIR spectra from 207-breast cancer patients using quantum cascade laser (QCL) imaging (5.55 mm-10.86 mm) and accurately differentiated between malignant and non-malignant stroma with 93.56% specicity. 5 Along with its numerous benets, MIR hyperspectral imaging also comes with limitations. It is currently limited to only imaging excised tissue. This is due to traditionally weak MIR light blackbody sources and the lack of optical bre that transmits sufficiently into the MIR spectral region. 6 The latter limitation has since been resolved since the advent of chalcogenide-based optical bres, derived from the chalcogen elements of group 16 in the periodic table (S, Se and Teexcluding O). 7 With the aim of providing point-of-care testing, in previous work, we developed low optical loss Ge 20 Sb 10 Se 70 atomic (at) % glass optical bres capable of transmitting MIR light from 2.5 mm-13 mm. 8 This glass system opens the possibility of utilising MIR hyperspectral imaging for the diagnosis of cancerous tissue, in vivo, through its incorporation into endoscopic probes. 9 The metalloid elements used in formulating these optical bres warrant an analysis of their cytotoxicity, prior to their application in point-of-care testing. Similar work was conducted by Wilhelm et al., 10 who investigated cell coverage by A549 human lung cells on glass preforms of Te 2 As 3 Se 5 (TAS) at%. This study reported inconsistent and non-reproducible cellattachment results, a problem attributed by the authors to the presence of arsenic (As) in the glass and a thin surface oxide layer. The toxic behaviour of the TAS glasses was then assessed by submerging the TAS bres in cell cultures and analysing the cell metabolic activity via a WST-1 colorimetric assay. It was concluded that, aer the TAS bres were washed in water for 24 hours, they produced no toxic effect towards the cells (95% to 103% cell viability), whereas unwashed bres produced a more toxic effect in comparison (65% to 69% cell viability).
As is included in many chalcogenide glass formulations due to its ability to stabilise the covalent glass network. However, the inclusion of this element into medical instruments (i.e., endoscopic probes) raises signicant concerns. This is because, even in micro-quantities, As is highly toxic and possesses the potential to irreversibly damage cells and tissue. 10 In previous work, 11 we replaced As with antimony (Sb) to produce a glass composition more suitable for biological applications. The toxicity of these Sb-containing compositions has however, not previously been evaluated. This forms the aim of this paper. To produce an in-depth evaluation on the cytotoxicity of Sbcontaining chalcogenide glasses and assess whether Sb-based compositions would be viable for their application into medical devices such as endoscopic probes.
Data related to the toxicity of antimony in humans come primarily from studies that investigate its connection with industrial-atmospheric exposure. Potkonjak and Pavlovich 12 examined the respiratory condition of 51 workers in a Yugoslavian antimony smelter with chest X-ray analysis. X-ray changes indicated antimoniosis. The workers were exposed to 17 mg m À3 to 86 mg m À3 antimony dust for 9 to 31 years. Antimony trioxide (Sb 2 O 3 ) made up 40% to 90% of the dust, antimony pentoxide (Sb 2 O 5 ) 2% to 8% and free silica 1% to 5%. Positive Xray ndings characterised antimoniosis as diffuse, densely distributed punctate opacities, irregular in shape, and with a diameter less than 1 mm in the lung. In addition, chronic coughing was reported in 60.8% of workers, chronic bronchitis in 37.2%, chronic emphysema in 34.5%, inactive tuberculosis in 18.2% and pleural adhesions in 27.3%.
The chemo-toxicological characteristics of arsenic and antimony are very similar. 13 In the trivalent state, both behave clastogenically in vivo and in vitro, are not directly mutagenic, and have a carcinogenic potential. 14,15 Genotoxicity of antimony has been shown to be valence dependant but the underlying mechanism for its genotoxicity remains unclear. 14, 16 Kuroda et al. 17 used the sister chromatid exchange (SCE) assay to show that Sb 2 O 3 at a concentration of 0.34 mL mL À1 induced signicant genotoxic activity, whereas Sb 2 O 5 at a concentration of 40 mL mL À1 did not. A biomonitoring study by Cavallo et al. 18 evaluated genotoxicity in 23 workers exposed to Sb 2 O 3 (via skin contact and the respiratory pathway) using SCE assay, micronucleus tests and (Fpg)-modied comet assay. The results suggested that Sb(III) toxicity results from oxidative DNA damage. It has also been suggested by Grosskopf et al. 16 that Sb(III) interferes with proteins involved in nucleotide excision repair, resulting in the partial retardation of this pathway and an indirect mechanism in the genotoxicity of Sb(III).
In view of this, we present a study that aims to evaluate the cytotoxicity of Ge 20 Sb 10 Se 70 at% glass optical bres on 3T3 mouse broblast cells. To observe the toxicity of the bres, 3T3 broblast proliferation was investigated. This involved two trials. Firstly, cells were subjected to elution trials, in which the effect of any glass-derived leachates on the proliferation of the cells was assessed. The concentration of each metalloid species (leachates) was measured, by exposing glass bres to deionised water over a 14 day period. This allowed for further evaluation of the cell viability outcome. This was followed by direct contact trials, which were performed to observe the adhesion/ attachment behaviour of cells on the surface of the bre. Cell viability was assessed using the neutral red assay; this chemosensitive assay was based on the ability of viable cells to incorporate and bind to the supravital neutral red dye. Since genotoxic activity is observed in literature as a response to Sb 2 O 3 , we generated oxides at the glass surface by exposing the Ge 20 -Sb 10 Se 70 at% glass optical bres to aqueous H 2 O 2 , (H 2 O 2(aq.) ) and the potential toxicity of the oxide produced on 3T3 broblasts was evaluated.

Materials and methods
2.1. Ge 20 Sb 10 Se 70 at% glass optical bre pre-treatments and sample codes Ge 20 Sb 10 Se 70 at% bulk glass rod was synthesised in high purity, using 99.999% purity elements, in a pre-puried silica ampoule, under a 10 À6 Pa vacuum at 900 C for 12 h. The bulk glass preform was then drawn into a bre with a diameter of 210 mm AE 20 mm. More details of the melting procedure can be found in the ESI. † The Ge 20 Sb 10 Se 70 at% bres were cleaved using a rubytipped scribe (S90R, Thorlabs) under ambient conditions into 5 mm and 10 mm lengths (for direct contact and elution trials respectively) and were subjected to one of three different post-bre-drawing pre-treatments, as indicated below. All the pretreatments were conducted under ambient conditions. 2.1.1. As-drawn (AD) and propylamine etched (PE) bre. As-drawn (AD) bre pieces were cleaved as described above, then washed in triplicate with acetone (Fisher Chemicals, 99.5%) and isopropanol (Fisher Chemicals, 99.5%). During each wash, the bres were placed in a 10 mL scintillation vial and held in an ultrasonic cleaner for 5 min. Following this, the bres were le in ambient conditions to allow the solvent to evaporate then wrapped in lens tissue (Ted Pella, Inc.) and stored in a silica glass vial for up to 5 days before use.
To prepare propylamine etched (PE) bres, as-drawn cleaved bre pieces were collectively submerged in 5 mL of propylamine (Sigma-Aldrich, 99.0%), within a scintillation vial for 0.5 h, then washed in triplicate with acetone, and isopropanol. During each wash, the bres were placed in a 10 mL scintillation vial and held in an ultrasonic cleaner for 5 min. Following this, the bres were le in ambient conditions to allow the solvent to evaporate, then placed into a new scintillation vial, wrapped in lens tissue, and transferred into an MBraun glove box, with a nitrogen atmosphere, with 0.6 ppm H 2 O and 0.1 ppm O 2 for storage up to 5 days before use.
2.1.2. H 2 O 2 aq. treated bres: O10, O30, and O60. Asdrawn, cleaved bre samples were initially triplicate washed in acetone and isopropanol then submerged in 30% w/v aqueous hydrogen peroxide (H 2 O 2(aq.) ) (Fisher Scientic) for 10, 30 and 60 min. Aer the timed submersion, the H 2 O 2(aq.) was decanted from the bres, that then were dried for 12 h under dust cover, in ambient conditions. Aer this process, a white, partially soluble glass surface oxide had formed. 19 2.1.3. H 2 O 2 aq. treatment followed by washing of bres: OW. As-drawn, cleaved bre samples initially underwent the same pre-treatment as the O60 bres (see Section 2.1.2). In the same way, a white oxide formed on all the bre surfaces. However, the bres then underwent further treatment as follows. Triplicate washing in acetone and isopropanol was carried out, while in an ultrasonic cleaner for 10 min (as described in Section 2.1.1), and this was observed to remove the white surface layer from the bres.
2.1.4. Further pre-treatment of bres. For all conditions mentioned above (see Sections 2.1.1 to 2.1.3), bres intended for direct contact trials were individually solvent washed, allowed to dry and wrapped in lens tissue to prevent ongoing physical damage to the bre surfaces, whereas bres intended for eluate production were collectively solvent washed, allowed to dry and wrapped in lens tissue.
Aer the pre-treatments described in Sections 2.1.1, 2.1.2 and 2.1.3, all bre samples intended for direct contact and elution trials were disinfected via ultraviolet (UV) sterilisation in a class II microbiological safety cabinet (Nuaire biological safety cabinets, Triple Red Ltd) for 90 min, under ambient atmosphere and temperature, before the trials began.
In order to prepare the elution medium (eluate) for each condition, bre samples from each condition were respectively submerged in culture medium (production method described in ESI †) for 24 h or 72 h, while being horizontally rotated in a sterile 20 mL disposable scintillation vial, at 50 rpm (revolutions per minute, Denley Spiramix 5). Oxidised samples (OW, O30 and O60) were only submerged for 24 h. Different submersion times were investigated to see what effect a longer bre-matrix interaction time would have on cell viability. Aer 24 h or 72 h, the culture medium, now the elution medium (eluate) was decanted from the bres and diluted. The following dilutions by volume of the eluate were then produced using the culture medium as the diluent: 100%, 75%, 50%, and 25%. The diluents were placed into 5 mL sterile universal tubes and stored in a chemical refrigerator (for a maximum of 3 days) at 4 C until needed, aer which the diluents were warmed to 37 C in a water bath and used.
The control sample used in this work was tissue culture plastic (TCP), i.e. an untreated, breless well of a 96-well plate (material: polystyrene, ThermoScientic).

Cytotoxicity tests
2.2.1. Elution. For the elution trials, the as-drawn, Ge 20 -Sb 10 Se 70 at% optical bres were cleaved and prepared to satisfy the ISO standard (10993-12:2012) of 6 cm 2 of any solid material, with a thickness less than 0.5 mm, per mL of liquid. 20 In this trial, a total of 31.05 mL of culture medium was used as the elution medium, with 186.3 cm 2 of bre surface area. Cells were seeded onto 5 wells for each of the seven different conditions (viz.: AD-24 h, AD-72 h, PE-24 h, PE-72 h, OW, O60 and O30) in a 96-well plate at a cell seeding density of 15 Â 10 3 cm À2 and incubated (Nuaire DH Autoow CO 2 Air-Jacketed Incubator) at 37 C with 5 vol% CO 2 , for 24 h, to allow cells to attach to the bottom of each well. Aer the 24 h incubation period, the appropriate eluate was added to each well as above described, and cell viability assessed via the neutral red assay at day-1 and day-3, aer the addition of the eluate (this counting of days, as e.g. day 1, excludes the prior 24 h incubation period). Control 3T3 broblast samples were grown on the same well plate, however using unmodied culture medium (described in ESI †).
2.2.2. Direct contact. Following the pre-treatments, indicated in the sub-sections of Section 2.1, seven bre pieces (each of 5 mm length) from each pre-treatment condition were selected and transferred into a new 96-well plate (1 bre piece per well), so that the sample number of bre pieces for each pretreatment condition was 5, giving 35 bre samples in all. Two additional pre-treated bre pieces were added to the 96-well plate per pre-treatment condition (i.e. for each of: AD, PE, OW, O10, O30, O60) totalling now 49 bre samples in all; the new bre pieces were added specically to enable their later removal directly for scanning electron microscopic (SEM) observation. These additional bres were added due to the destructive nature of the neutral red dye; cells that have absorbed the dye, rupture during the destaining process 21 and so, in order to observe the intrinsic growth pattern of the cells, these two additional bres per pre-treatment condition were used, which would not undergo neutral red staining. Cells were then seeded onto all 49 bres in the 96-well plate at a seeding density of 12.4 Â 10 3 cm À2 . Cell viability was assessed via the neutral red assay, for those bre samples not destined for SEM observation, at day 1, 3, 7 and 14. Cell viability associated with each type of bre pre-treatment condition at each time point was also compared to the control samples (n ¼ 5) grown on TCP.
2.2.3. Cell viability aer direct contact and elution trials via the neutral red assay. To quantify the cell viability aer direct cell contact with the bres and the elution trials, the neutral red assay was performed. The method for this was as follows. 20 mL of neutral red stock was prepared in a 4 mg to 1 mL ratio of neutral red dye to deionised water in a 25 mL sterile universal tube (Sterilin, UK), and then protected from photocatalytic degradation by wrapping in aluminium foil (Terinex). Neutral red medium was then prepared in a 1 : 100 (neutral red stock : culture media) dilution under sterile conditions, centrifuged at 1200 rpm for 4 minutes so that any crystals that may have formed, would settle at the base of the 25 mL sterile universal tube; care was taken not to shake the tube, so as not to re-suspend the crystals. The neutral red medium was not stored before/aer an experiment. Any excess was discarded.
At the indicated time points (day-1, day-3, day-7, and day-14 for direct contact trials and day-1 and day-3 for elution trials), the cells were submerged in neutral red medium and incubated for 2 hours at 37 C with 5 vol% CO 2 . Aer incubation, the unincorporated neutral red was decanted and discarded. The cells were then washed in triplicate with phosphate buffer saline (PBS, Oxoid UK). Following the triplicate washes with PBS, a destaining solution was added to each well, to extract the incorporated dye from the viable cells. The destaining solution was prepared by volume in ambient temperature, by mixing pure ethanol (Scientic Laboratory Supplies, 99%), with deionised water and aqueous glacial acetic acid (99.74%, Fisher Scientic, UK)) to a ratio of 5 : 4.9 : 0.1, respectively. The 96-well plate was then placed on a plate shaker (Titramax 100, Heidolph) at 300 rpm for 10 min under ambient conditions to extract the incorporated neutral red dye from the cells. Finally, the 96-well plate was placed in the BioTek™ ELx800™ UV (Fisher Scientic) plate reader (primary wavelength 490 nm, reference wavelength 630 nm), to quantify the neutral red dye concentration. Cell viability was expressed as a percentage of the control of which, a cell viability under 70% of the viability expressed by the control cells, indicated material cytotoxicity, as stated by the ISO standards ISO 10993-5:2009. 22 2.2.4. Chemical analysis: Ge, Sb, and Se concentration in leachates -viainductively coupled plasma mass spectrometry (ICP-MS). Ge, Sb, and Se concentrations (mg L À1 ) in aqueous leachates, aer exposure to AD, PE and O60 were measured using the single quadrupole inductively coupled plasma mass spectrometer (ICP-MS) iCAPQ. For these leaching experiments, bre samples from each condition were respectively submerged into deionised water at 37 C for 1, 3, 7 or 14 days, while being horizontally rotated in a sterile 20 mL disposable scintillation vial, at 50 rpm (revolutions per minute, Denley Spiramix 5).

Materials and methods (ESI †)
See ESI for the following details: (1) Bulk Ge 20 Sb 10 Se 70 at% preform fabrication and bre preparation.
(2) A physical analysis of Ge 20 Sb 10 Se 70 at% glass optical bres using coherence scanning interferometry and scanning electron microscopy.
(3) A pH physical analysis of the following eluates: AD, PE and O60.
(4) The 3T3 broblast cell-line and method for producing the culture medium.
(5) Statistical analysis method performed on the results of this work.

Results
To recap, to simplify interpretation of the results, as described in Section 2.1, AD, PE, OW, O10, O30 and O60 bre samples were prepared. Fibres were initially cleaved into 5 mm and 10 mm long lengths (for direct contact and elution trials, respectively), then subjected to one of three pre-treatment conditions. These were, submersion in propylamine for 0.5 h (PE); submersion in H 2 O 2(aq.) for 10, 30 and 60 min and subsequent drying in ambient conditions for 12 h, to allow the build-up of a glass oxide layer (O10, O30, O60, respectively) and submersion in H 2 O 2(aq.) for 60 min and subsequent drying as previously described and triplicate washing in both acetone and isopropanol to remove the oxide layer (OW) then drying in ambient conditions via evaporation. AD bres were not subjected to any pre-treatment conditions.

Chemical analysis: Ge, Sb and Se concentration in aqueous leachates
As in Fig. 1 (37 C leaching temperature), O60 resulted in the highest Ge, Se, and Sb concentration, compared to AD and PE leachates. The concentration of Se and Sb in O60 dropped as the time for leaching increased, from 3858 mg L À1 and 2702 mg L À1 , respectively in D1 to 1207 mg L À1 and 459 mg L À1 , respectively, at D14. This drop is, however, more notable from D7 to D14, a 175.6% (Se) and 364.8% (Sb) drop, compared to the drop in concentration from D1 to D7 (Se: 16% and Sb: 26.7%). This time dependant relationship with concentration is also observed in AD and PE leachates. In AD, Ge and Sb concentrations increase by 222% and 159%, respectively, from D1 (Ge: 0.31 mg L À1 , Sb: 0.19) to D14 (Ge: 1.02 mg L À1 , Sb: 0.50 mg L À1 ), while Se is seen to decrease by 61.7% from D1 (Se: 0.12 mg L À1 ) to D14 (Se: 0.045 mg L À1 ). In PE, Ge and Sb concentrations increase by 1955% and 1757% from D1 (Ge: 0.13 mg L À1 , Sb: 0.08 mg L À1 ) to D14 (Ge: 2.63 mg L À1 , Sb: 1.48 mg L À1 ), however, this increase is more prominent from D7 to D14 (Ge: 832% and Sb: 663%), compared to D1 to D7 (Ge: 121% and Sb 143%). Se was observed to increase from 0.017 mg L À1 , in D1, to 0.052 mg L À1 in D14 (a 196.5% increase).

Cell response
3.2.1. Elution. Fibre samples were prepared for elution trials by satisfying the ISO standard (10993-12:2012) of surface area of 6 cm 2 per mL of liquid for any solid material with a thickness below 0.5 mm, 20 see Section 2.2.1. To prepare the eluate for each pre-treatment condition, bres from each condition were separately submerged together in culture medium for 24 h or 72 h, while being horizontally rotated in a scintillation vial at 50 rpm.
Aer day-1 (from Fig. 2A), broblasts show a viability of >80%, in all eluates, excluding O30 and 060 which only show a viability of <20%. In Fig. 2A, a lower eluate concentration in PE-24 h and PE-72 h is seen to result in higher broblast viability. This correlation is also observed in AD-24 h, AD-72 h, and OW, up to an eluate concentration of 50%. At an eluate concentration of 25% in AD-24 h, AD-72 h and OW, the cell viability dropped by 10.4%, 6.8% and 4%, respectively. No statistically signicant differences were found between the viabilities of cells grown in AD-24 h, AD-72 h, PE-24 h, PE-72 h and OW. Signicant differences were found between the viabilities of cells grown in O60 and O30 and all other eluates, but not between O60 and O30. Fig. 2B shows that cell viability decreased in all eluates, excluding 50% and 25% concentrated PE-24 h, which increased by 8.2% and 10.7%, respectively, when compared to the cell viability observed in PE-24 h aer day-1. Cell viability was highest in 25% concentrated PE-24 h (111.1% AE 3.1%) and lowest in 100% concentrated PE-72 h (53.9% AE 5.9%). A notable drop in the pH of O60 was also observed. The pH of the growth media was originally 7.42, and this dropped to 2.44 aer 24 h interaction time with O60 bres. It is inferred that this occurred when glass oxide from the surface of the O60 bre samples dissolved into the growth media.
3.2.2. Direct contact. Fig. 3 shows the viability of broblasts over a period of 14 days grown on AD, PE, OW, O60, O30 and O10 bre surfaces, following direct contact trials as described in Section 2.2.2. Most notable is the signicantly lower cell viability on O10, O30 and O60 bre surfaces from day Fig. 1 The concentration of Ge, Sb and Se in aqueous leachates after exposure of fibre (conditioned as AD, PE and O60) to deionised water at 37 C for 1, 3, 7 and 14 days (D1, D3, D7 and D14, respectively).  1-14. On day-1, cell viability was highest on PE bre surfaces (139.9% AE 6.1%) and lowest, (not including O10, O30 and O60 bres) on OW bre surfaces (68.7% AE 6.2%). Also, on day-1, AD surfaces resulted in a broblast cell viability of 123.4% AE 15.2%. On day-3, cell viability on AD and PE bre surfaces decreased to 92.8% AE 8.2% and 110.2% AE 2.2%, respectively. The broblast cell viability on the OW bre surface, in contrast, had increased by 29.74% to reach 98.5% AE 6.7%. On day-7, a small increase in cell viability on AD (6.6% increase) and OW (12.5% increase) surfaces was recorded rising to viz.: 103.4% AE 4.7% and 111.0% AE 14.3%, respectively. A 2.4% decrease in cell viability was recorded on PE surfaces. On day-14, cell viability on AD, PE and OW surfaces had decreased by 5.0%, 20.3% and 15.3% relative to their 7 day viabilities, respectively.

SEM analysis
3.3.1. Day-1. On day-1, a very low distribution of broblasts was observed on AD, PE and OW surfaces, as seen in Fig. 4A1, B1 and C1, respectively. Cell shape on day-1 ranged from elongated structures with lamellipodia protruding from the surface (Fig. 4A1 and B1) to globular structures possessing no observable lamellipodia (Fig. 4C1). No cells were observed to have adhered on O10, O30 and O60 surfaces at day-1, as seen in Fig. 4D1, E1 and F1. A single cell was however seen to have attached onto the O10 surface (see inset in Fig. 4D1), maintaining a globular structure.
3.3.2. Day-3. On day-3, a very low distribution of broblasts was seen on AD and OW surfaces ( Fig. 4A2 and C2). A higher distribution of broblasts on the PE surface was observed (Fig. 4B2). Cell morphology on PE and OW surfaces ranged from fully spread to globular, whereas only globular cells are observed on the AD surface. Although no extensive attachment of cells was observed on O10, O30 and O60 surfaces (Fig. 4D2,  E2 and F2), some structures assumed to be broblasts were found still attached to O10 and O30 surfaces (see insets in Fig. 4D2 and E2).
3.3.3. Day-7. By day-7, cell conuency is observed on AD, PE and OW surfaces, as seen in Fig. 4A3, B3 and C3, respectively. Areas on Fig. 4A3, B3 and C3 indicated by orange brackets show the section of bre that was in contact with the well plate, hence the reduced cell density. No cells were observed on O60, O30 and O10 bre surfaces in contrast to day-1 and day-3. In Fig. 4D3, indicated by the inset image, are structures found on the bre surface, not believed to be cells. Due to their crystalline appearance, it is believed that these structures are inorganic crystals developed from interaction of the culture medium with the oxide surface of the bre. (1) Surface roughness of the bres obtained via coherence scanning interferometry.
(4) Concentration of Ge, Sb, and Se obtained via intercoupled plasma mass spectrometry.

Discussion
In this study, a simple assessment of the cytotoxicity of Ge 20 -Sb 10 Se 70 at% chalcogenide glass optical bre is presented. AD bres were tested as they represent optical bres that could be applied to MIR-based medical devices (i.e. endoscopic probes). When exposed to air, a passivating oxide layer begins to form on the surface of chalcogenide-based bres. Fibres exposed to air for long periods of time (e.g. >1 year) can develop a surface oxide layer, with a thickness >30 A. 19 It has been previously shown 10 that, when exposed to air for an extended period of time (3 years), bres exert a signicant toxic effect, however, when these bred are washed, this toxic effect disappears. PE bres are a representation of these "washed" bres. When bres were exposed to propylamine, this oxide layer was completely stripped off the bre, and cells are therefore, exposed to 'bare' glass and not an oxide layer as in AD.
The bres used in this study were exposed to air for no longer than 90 days, and so, an oxide layer thick enough to inuence cell response, it is suggested, would not have developed. To observe a similar glass-oxide based cytotoxicity as in ref. 10, a surface oxide layer was forced to develop via H 2 O 2 aq. exposure. Finally, the OW treatment was selected to observe any remnant cytotoxicity in bres, post H 2 O 2 aq. treatment. In this condition, cells were exposed to bres that were rst treated with H 2 O 2 aq. then washed to remove the surface oxide layer.
The surface oxide layer in O10-O60 was observed to be partially soluble in the cell growth medium. Once dissolved, this resulted in a notable drop in pH from the physiological 7. 49 (ref. 23) to the acidic 2.44. This drop in pH was shown, through direct contact and elution trials, to have a toxic effect on cell viability. The elution trials show that a 1 day exposure to oxidised eluates was enough to result in a cell viability of <10% (see Fig. 2). Direct contact trials also show that a 1 day exposure to bre surfaces with a glass oxide layer results in a cell viability of <30%, which dropped to <3% aer day-3 (see Fig. 3).
This toxic effect on broblast cells may be explained as follows. Alterations to the cellular microenvironment can signicantly affect cell physiology and induce pathology. 24 An important aspect of the extracellular environment is its pH. This must be kept within strict boundaries (between 6 and 8 according to ref. 23) to facilitate correct cellular function and prevent cell death. 25,26 Pathologies associated with an acidic extracellular environment include ischemia (pH <6.3 (ref. 27)), hypercapnia 28 and metabolic acidosis (pH <7. 35 (ref. 29)). Furthermore, mitochondrial dysfunction and the acidication of the extracellular environment have been linked to increased levels of reactive oxygen species (ROS) within cells. 30,31 ROS (such as, superoxide (O 2 À ), singlet oxygen ( 1 O 2 ), hydroxyl radical (OH . ) and H 2 O 2 ) are formed as products under physiological conditions as a result of the partial reduction of O 2 . 32 Low concentrations of ROS are required to maintain normal physiological functions such as cellular proliferation, signal transduction, immunity, and genetic expression. 32,33 However, an increased concentration of ROS results in the induction of oxidative stress. 24 In a study by Teixeira et al. 24 HEK293 cells were used to evaluate the effects of extracellular acidication by ambient temperature HCl aq. addition. They found that lowering the extracellular pH from 7.2 to 5.8, lowered cell viability by 70%, decreased cytosolic pH, hyperpolarised the mitochondrial membrane potential and increased the ROS levels. These results suggest that an acidic extracellular environment can induce cell death through an mPTP opening and ROS-mediated pathogenic pathway. This supports earlier work by Xue and Lucocq 34 who showed that low extracellular pH results in the activation of the c-Jun N-terminal kinase (JNK) pathway, an important regulator of various cellular processes, including apoptosis, in Swiss 3T3 cells. 35,36 In our work, the extracellular pH dropped from 7.49 to 2.44, and this was paralleled with a +70% drop in cell viability observed in both elution and direct contact trials aer 24 h. Considering former work, 24 the primary cause of cell death observed in our work is suggested to result from a potential increase in the concentration of ROS within the O10, O30 and O60 eluates.
A secondary potential cause of cytotoxicity involves the possible inuence of Ge, Sb and Se metalloid oxides. Metal oxide nanoparticles (MO-NPs) are known to be highly toxic to biological tissue, so much so that some authors have demonstrated their potential applicability in cancer therapy. [25][26][27][37][38][39] These oxide nanoparticles have shown toxic effect via proapoptotic activity, autophagy, proliferation inhibition, metal ion release and the increase of ROS. 38,40 In a study by Wilhelm et al. 10 on the biocompatibility of Te-As-Se glass bres, they observed pronounced toxicity only for bres exposed to air for 3 years. This toxicity was attributed to the formation of a soluble As 2 O 3 layer on the bre surface, which when exposed to aqueous solution, results in the release of arsenic (As) into solution as both As(III) and As(V). As then exert its toxicity by inactivating vital enzymes involved in ATP generation and DNA synthesis and repair. 41 Similar results were observed in this study when bre surface oxidation was chemically enforced over a shorter time. The proliferation rates of broblasts exposed to O10, O30 and O60 surfaces and eluates were signicantly lower, when compared to the other glass surfaces. As mentioned above, an additional explanation of this is the release of metal ions into the growth media. The question to now address is, which metal oxide is the most plausible cause for this toxicity, GeO 2 , Sb 2 O 3 , Sb 2 O 5 , Se-O or SeO 2 ?
It is generally accepted, that germanium metal nanomaterials have low toxicity, 42 are non-carcinogenic and even inhibitory to the formation of tumour cells. 43,44 Conversely, as discussed by Lin et al., 45 inorganic germanium compounds such as GeO 2 , can result in neuro-or nephrotoxicity, but only aer very high doses (e.g. a minimum oral uptake of 5 g GeO 2 per day for a 70 kg adult) and long-term exposures. In the same paper the authors showed that fabricated nano-Ge and nano-GeO 2 are non-toxic to Chinese hamster ovary (CHO) K1 cells, displaying a relative viability greater than 70%. This shows that Ge and its products possess very low toxicity, such that it has found use as a health supplement, in the form of Ge-132. 46 The levels of Ge observed in our work were considerably lower than the toxic levels indicated in literature (see above). While assessing the concentration of metalloid elements in our leachates via ICP-MS, we observed a maximum Ge concentration of 823.8 mg L À1 , in O60 leachates aer 3 days of dissolution, the toxicity of which, is attributed to the acidic extracellular microenvironment, generated by the drop in pH. We also observed a maximum Ge concentration of 1.02 mg L À1 and 2.63 mg L À1 , in AD and PE leachates respectively, aer 14 days of dissolution, and no toxicity was observed. GeO 2 is therefore assumed not to contribute to cell death observed in our work.
Regarding antimony, a weak congruency exists amongst investigators on the toxicity of Sb 2 O 3 . Titma et al. 47 have shown that, when human lung epithelial (A549) cells are exposed to 100 mg mL À1 of Sb 2 O 3 , cell viability falls from approximately (approx.) 170% aer 12 hours, to approx. 30% aer 36 hours. This agrees with the conclusions drawn by Mann et al. 48 and Verdugo et al. 49 that Sb 2 O 3 , Sb(III) and Sb(V), inhibit cell growth and induce apoptosis, by utilising similar signalling pathways to As 2 O 3 , As(III) and As(V), resulting in similar toxicities in acute promyelocytic leukaemia (NB4 APL) and human embryonic kidney (HEK-293) cell lines. On the contrary, Omura et al. 50 showed a lack of toxicity resulting from Sb 2 O 3 on the count, motility and morphology of rat and mouse sperm cells. Even at a repetitive administration of a 1.2 g kg À1 dose for 4 weeks, no toxicity was reported in mouse or rat testes. A possible explanation for this discrepancy is that the tissue distribution of Sb 2 O 3 does not involve the testes, but primarily, the liver and kidneys. To add to the incongruity, Bregoli et al., 51 within the same study, showed that Sb 2 O 3 nanoparticles (NPs) demonstrated toxicity towards primary cultures of human hematopoietic progenitor cells but not to human hematopoietic immortalised cell lines. The authors assigned this difference to the tumorigenic alterations of cells lines. In our studies of enforced surface oxidation of Ge-Sb-Se glass optical bres, we cannot rule out the presence also of antimony pentoxide, however we nd no toxicology reports about this particular oxide. As discussed in ref. 52, the human body is primarily exposed to antimony, via the inhalation route. Exposure via the gastrointestinal (GI) tract is low (<1%) and is limited by the emetic properties of antimony compounds. Finally, uptake through the skin makes no signicant contribution to systemic exposure. 52 The occupational aerosol exposure limit set by the American Conference of Governmental Industrial Hygienists (ACGIH) is 0.5 mg m À3 . 53 Cases where exposures are signicantly in excess of this (4 mg m À3 to 86 mg m À3 ), 12,14 have been associated with pulmonary toxicity. 12,14,53 Pustular skin eruptions and antimony dermatosis have also been observed in people working with antimony and antimony salt. 12,14,54 In our work, Sb ion concentration (see Fig. 1) in AD leachates aer leaching in deionised water for 24 h and 72 h elution, at 37 C (0.194 mg L À1 (i.e. 194.4 mg m À3 ) and 0.165 mg L À1 (i.e. 165.2 mg m À3 ), respectively) far exceeded the limits set by the ACGIH (0.5 mg m À3 ). Although elution was in aqueous media, and not in pure deionised water, it is surmised that the eluant Sb concentration would very likely have mirrored the aqueous leachate Sb concentration. It is notable that no cytotoxicity was observed in cells exposed to these eluates. These results seem to contradict Titma et al., 47 who showed that 100 mg mL À1 of Sb 2 O 3 NPs result in a drop in cell viability from 100% to $45% in Balb/ c 3T3 cells. These discrepancies are discussed later in this Discussion.
Selenium (Se) is an essential trace element incorporated into selenoproteins as selenocysteine. There are 25 selenium-based proteins in the human selenoproteome, with functions ranging from protection against oxidative damage to the regulation of cellular processes. 14 Conversely, Se compounds have been shown to possess a clear cytotoxic activity against malignant cells. 55,56 Forootanfar et al. 57 investigated the cytotoxicity of SeO 2 and Se NPs (spherical and ranging from 80 nm to 220 nm in size), using the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide) assay. They found that the IC 50 was 6.7 mg mL À1 AE 0.8 mg mL À1 in the MCF-7 cell-line treated with SeO 2 . Se NPs produced the same effect but at a signicantly higher concentration of 41.5 mg mL À1 AE 0.9 mg mL À1 . In previous work by Suzuki et al., 55 it was concluded that Se compounds induced apoptosis by causing endoplastic reticulum stress and activating the intrinsic apoptotic pathway. We however, observed no cytotoxic effect on cells exposed to AD eluates, containing 0.05 mg L À1 (aer a 72 h elution time), see Fig. 1.
A tentative explanation for the discrepancies with the literature discussed here lie in the action of Se in high oxidative stress situations, and its interaction with As (and by extension, Sb). Firstly, it has previously been shown 58 that Se-NPs activate the ATF4, SOD2 and Bcl-xL genes (the anti-oxidative stress pathway), to counter hydroquinone (HQ) induced oxidative stress and protect human dermal broblast cells. Additionally, internal ROS species were determined to be lower per cell, in cases where cells were exposed to Se-NPs (during HQ induced stress) in comparison to no exposure. Secondly, some authors have found that Se decreases the toxicity of As. 59,60 As and Se have similar methylation pathways, and so, can mutually impede the secretion of their methylation metabolite. 61 This led to the hypothesis that an As-Se compound is formed, that causes less damage on cells than As or Se alone. 62 This compound was rst discovered as seleno Bis(S-glutathionyl) arsinium ion [(GS) 2 AsSe] À . 63 It was later found that simultaneous entry of As and Se into a cell, results in the formation of (GS) 2 AsOH rst, which is then attacked by HSe À , displacing its -OH group, to form [(GS) 2 AsSe] À , 64 which can later be excreted from the cell. Furthermore, Rossman and Uddin, 65 proposed that Se can guard against As induced oxidative damage through the upregulation of selenoproteins, thioredoxin reductase and glutathione peroxidase. Due to the similar chemical and toxicological characteristics of As and Sb, we suggest that a similar interaction between Se and Sb, forming a seleno-antimony (Sb-Se) detoxication conjugate may have occurred in cells exposed to AD and PE eluates, and so, resulted in the mutual reduction of the toxicity of Sb and Se, and provided increased protection from Sb or Se based oxidative stress. This hypothesis is supported by Feng et al., 66 who showed that Se alleviated the toxicity of Sb in plants by inhibiting the uptake of Sb and relieving oxidative stress derived from Sb exposure. 67 The level of Se generated in this study (discounting the levels produced in O60 leachates) is far lower than the permissible exposure limit of Se (400 mg day À1 ). 68 The level of Sb, produced in this study (discounting levels produced in O60 leachates) is however, higher than the permissible exposure limit set by the ACGIH (0.5 mg m À3 ). 53 We have however shown that, even when this is the case, there is a lack of cytotoxicity observed from AD bres (via cell adhesion and cell propagation). This demonstrates the potential biocompatibility of medical devices developed using the Ge 20 Sb 10 Se 70 at% glass composition.
Here, the surface texture of Ge 20 Sb 10 Se 70 at% optical bres was modied by etching with propylamine under ambient conditions or through oxidation by H 2 O 2(aq.) (see Section 2.1). The results (shown in Fig. S1 and S2 of ESI †) have demonstrated that Ge 20 Sb 10 Se 70 at% optical bres are more susceptible to surface damage caused by oxidation by H 2 O 2(aq.) for 60 min, than propylamine etching, as surfaces that underwent the former, generated the rougher surfaces (PE Sa ¼ 169 nm AE 15 nm and O60 Sa ¼ 236 nm AE 84 nm), results not shown here. This is not believed to have inuenced cell viability, as cells would not have had the time to respond to the altered rougher surface due to the drastically altered chemistry of the cellular microenvironment aer exposure to oxidised eluates and bre surfaces.

Conclusions
Using a 3T3 broblast cell model, we provide evidence that unmodied Ge 20 Sb 10 Se 70 at% optical bres do not produce a cytotoxic response. A cytotoxic response was only observed aer a glass oxide layer was forced to develop on unmodied, as-annealed, Ge 20 Sb 10 Se 70 at% optical bres, through submersion in H 2 O 2(aq.) . Such oxide growth does not occur under normal ambient conditions, nor on exposure to pH neutral water, but on exposure to a strong oxidising agent like H 2 O 2(aq.) . The primary cause of toxicity is attributed to the dramatic pH change that occurred in the extracellular environment, aer interaction with the glass oxide. This drop in pH from the physiological 7.49 to the acidic 2.44 resulted in a +70% decrease in cell viability aer 24 h in both elution and direct contact trials. A secondary negative inuence on cell viability is attributed to the presence of Sb 2 O 3 and SeO 2 in solution. The lack of cytotoxicity in cells exposed to AD and PE samples is attributed to the mutual antagonistic effects of Sb and Se. The lack of cytotoxicity from the unmodied bres in our work, demonstrates the potential biocompatibility of medical devices developed using the Ge 20 Sb 10 Se 70 at% glass composition.

Conflicts of interest
There are no conicts to declare.