Abstract
Purpose:
This study assessed the cytotoxicity of the impurities detected in the perfluorooctane (PFO) batches for vitreoretinal surgery that were associated with serious adverse incidents of ocular toxicity, namely, the perfluorooctanoic acid (PFOA), 1H,1H,7H-dodecafluoro-1-heptanol (DFH), 1H-perfluorooctane (1H-PFO), ethylbenzene, anhydrous p-xylene, and perfluoro-2-butyltetrahydrofurane, and two additional substances 1H,1H,1H,2H,2H-perfluorooctane (5H-PFO) and hexafluoro-1,2,3,4-tetrachlorobutane.
Methods:
Serial dilutions were tested by in vitro direct contact cytotoxicity test, validated in accordance with the ISO 10993-5:2009 standard using BALB3T3 and ARPE-19 cell lines, after sample application for 24 hours.
Results:
Six of the eight tested substances were cytotoxic according to the above-mentioned ISO standard. Anhydrous p-xylene, ethylbenzene, and PFOA were the most cytotoxic impurities as traces 1.55 ppm, 1.06 ppm, and 28.4 ppm reached the cytotoxicity limit, respectively. Hexafluoro-1,2,3,4-tetrachlorobutane, DFH, and 1H-PFO were cytotoxic at 980, 22,500, and 123,000 ppm, respectively. Both 5H-PFO and perfluoro-2-butyltetrahydrofuran were non-cytotoxic at the highest available concentrations (≥970,000 ppm). The dose-response curves allowed to calculate the cytotoxic concentration (CC30) for each tested substance that would reduce 30% of cell viability and corresponding to the cytotoxicity threshold according to ISO 10993-5.
Conclusions:
Our study determined the in vitro cytotoxicity of several impurities in PFO associated with serious adverse incidents in retinal surgery patients.
Translational Relevance:
Severe cytotoxicity of some impurities previously found in toxic perfluorocarbon liquids was confirmed. The cytotoxicity test validated according to the ISO 10993-5:2009 standard is a sensible and fast method for reliable detection of the cytotoxicity in perfluorocarbon liquids to guarantee maximal safety for the patients.
At least six optical density (OD) values at 570 nm were acquired at each 96-well microplate for all samples (impurity, vehicle, positive control, and negative control) using the Absorbance Microplate Reader ELx 80 (Biotek Instruments, Winooski, VT, USA).
Mean percentage of reduction of cell viability was calculated as follows:
\begin{eqnarray*}\!\begin{array}{l}
\begin{array}{@{}l@{}}
{\rm Mean}\; \%\; {\rm of\; reduction\; of\; cell\; viability} = {\rm mean} \frac{{\rm OD\; tested\; sample} \,-\, {\rm OD\; blank}\;\%} {{\rm OD\; vehicle} \,-\, {\rm OD\; blank}}\\ \left({\rm blank} = {\rm vehicle\; with\; no\; cells} + {\rm MMT\; or\; NRU} + {\rm dissolving\; agent} \right) \end{array}\end{array}
\end{eqnarray*}
Mean percentage of reduction of cell viability of ARPE-19 and BALB3T3 cells and standard error of the mean (SEM) were calculated for each concentration of tested substances and both controls.
Theoretical values of cytotoxic concentration (CC
30), corresponding to the concentrations that would reduce 30% of cell viability in ARPE-19 and BALB3T3 cells, for substances including
p-xylene, ethylbenzene, PFOA, DFH, hexafluoro-1,2,3,4-tetrachlorobutane, 1-HPFO, 5H-PFO, and perfluoro-2-butyltetrahydrofuran were calculated from the fitted regressions reported in the
Figure. The differences between cell lines in percent mortality were determined by using Student's
t-test.
It is now acknowledged that the toxicity induced by AlaOcta PFO batches was related to the chemical damage and direct contact of the toxic impurities with the retina, causing retinal irreversible damage.
9 The company was clearly negligent in their responsibility for selling untested AlaOcta PFO toxic batches. The risk of presence or appearance of potentially toxic substances in PFCL-based medical devices used as vitreous tamponades needs to be excluded by the manufacturers using accurate analytical methods and adequate safety criteria for their clinical use.
6-8,11,18,21
The different physical-chemical methods such as gas chromatography coupled with mass spectroscopy, nuclear magnetic resonance (NMR), ultraviolet-spectrophotometry and ion selective potentiometry
10,13,21 may be used to detect and quantify the presence of specific potentially hazardous substances. However, these methods do not provide any information on their toxicity. According to the ISO standard
13 the concentration of the contaminants in PFCLs should be as low as possible; however, the acceptance limits for specific PFCLs impurities are not clearly described.
In this study the substances that were previously found as contaminants of the toxic batches of AlaOcta PFO
8,11,17 along with those of 5H-PFO and hexafluoro-1,2,3,4-tetrachlorobutan were tested by in vitro validated direct contact cytotoxicity test
15 to obtain dose-response curves. The dose-response curves allowed to the level of the cytotoxicity of each tested substance based on an estimated CC
30 (cytotoxic concentration that would reduce cell viability of 30%) and the lowest tested concentration that reached or passed the cytotoxicity threshold as defined by the ISO 10993-5
12 to be determined. The CC
30 value allowed the classification of tested substances on the basis of the level of cytotoxicity. The calculation of CC
30 was based on the dose-response curves that were obtained experimentally by application of seven to 11 concentrations of each tested substances to two cell lines and the quantification of reduction of cell viability after 24 hours of application. The CC
30 values calculated according to the equations confirmed the order of cytotoxicity obtained experimentally for all tested substances. The CC
30 values calculated for ethylbenzene and
p-xylene were similar and both very low, corresponding to 1.06 ppm and 1.21 ppm, respectively, indicating that both substances show very high cytotoxicity at ppm level in tested cell lines. Our findings showed that, among all tested substances, ethylbenzene and
p-xylene were the most toxic, followed by PFOA and DFH; hexafluoro-1,2,3,4-tetrachlorobutane and 1H-PFO were less but still cytotoxic, and finally, 5H-PFO and perfluoro-2-buthyltetrahydrofuran were definitely noncytotoxic, even when tested at the highest commercially available concentrations.
The toxicity of PFCLs is often caused by incomplete fluorination of hydrocarbons, the major impurities in PFCLs contain residual hydrogen-containing compounds and unsaturated carbon bonds.
20-22 Chang et al.
7 determined hydrogen content in PFO by NMR spectroscopy. Because the safety threshold of protonated impurity in PFCL for intraocular use was unknown, they sought to obtain liquid of the highest grade corresponding to protonated impurity content < 0.1 ppm and equivalent to the detection limit of the method.
16 The work of Sparrow et al.,
22 demonstrated the toxicity of 1H-PFO at the perfluorocarbon-fluid interface in tissue culture by a qualitative assessment. These studies
7,22 indicated the level of impurities that were used for approval of perfluoro-n-octane from the U.S. Food and Drug Administration.
Menz et al.
10 used physical-chemical determination of partially hydrogenated perfluoroalkanes through the ion-selective potentiometry after digestion of perfluorocarbon liquid.
13,21 This analysis determines the so-called H-value, defined as the ppm content of reactive partially hydrogenated perfluoroalkanes and to which an H-value <10 ppm was attributed as the safety threshold by Menz et al.
10 However, the ion-selective potentiometry does not detect toxic substances without a specific hydrogen-fluoride-containing compounds such as the hydroxyl impurities identified in the toxic batches of PFCL.
8,11,17
Our study showed that in vitro cytotoxicity test can provide direct information on cytotoxicity of impurities contained in PFCLs, independently from substances identification and quantification.
The findings in this study are in agreement with the results obtained by Ruzza et al.,
18 who studied the extent of ARPE-19 and BALB3T3 cell mortality after the application of 1H-PFO, PFOA, and 5H-PFO at three of the concentrations tested in this study and Romano et al.’s.
15 The toxicity threshold assessed for PFOA and DFH in this study only slightly differed from those evaluated by Srivastava et al.,
17 who found that PFOA at 0.06 mM was close to the toxicity limit, which corresponded to 25 ppm versus CC
30 of 17 ppm determined in our study, and DFH was toxic at 4.48 mM (corresponding to 1488 ppm vs. 1173 ppm in our study). It is possible that the small differences between the two studies were related to the different testing conditions (e.g., sample size, contact time and number of replicates).
Consistent with the findings from other studies,
14,15,17,21 the extent of mortality of ARPE-19 and BALB3T3 cells in this study generally increased at each increasing concentration of all tested substances.
With respect to analyses of the contaminant profiles of the toxic AlaOcta batches conducted by Menz et al.,
16 we might affirm that the toxicity of those batches could not be solely related to the presence of 1H-PFO or DFH, because their concentrations, which differed between the 59 and 875 ppm for 1H-PFO and between 29 and 45 ppm for DFH, were well below the cytotoxicity limit (CC
30) calculated in this study (CC
30: 55712 ppm for 1H-PFO and 1173 ppm for DFH). Instead, the PFO batches containing PFOA at concentrations varying between 50 and 700 ppm and a mix of
p-xylene isomers/ethylbenzene (82%:18% ratio) at 5 to 30 ppm
16 were surely toxic, because the quantity of these compounds was well above the cytotoxic limit found in this study (CC
30: 17 ppm for PFOA; 1.21 ppm for
p-xylene and 1.06 ppm for ethylbenzene). In addition, Srivastava et al.
17 also analyzed the contaminants in the toxic AlaOcta batches and found that DFH was contained at nontoxic concentration, whereas PFOA was detected at concentrations that reduced the ARPE-19 cell viability of > 70% and thus were inevitably cytotoxic. However, as also affirmed by the two groups of investigators,
16,17 the ocular toxicity of those AlaOcta batches were likely to be due to the combined effect of all contaminants rather than one specific substance; furthermore, the individual risk of each type of contaminant may be increased by the interaction with other contaminant groups.
16 In the presence of unknown contaminants, the cytotoxicity test represents a relatively fast and simple quality control method because it detects the overall cytotoxicity of samples and can be used for monitoring of the cytotoxicity during manufacturing processes.
Similarly to the previous study,
15 close contact between the contaminant diluted in PFO and the cell layer for 24 hour was obtained, and the cell mortality induced by cytotoxic substances was mainly located in the contact area under the PFO bubble.
In addition to murine fibroblast cells BALB3T3 cell lines required by the ISO 10993-5 standard,
12 we used the human retina-derived ARPE-19 to imitate more closely what happens during vitreoretinal surgery. In line with previous studies,
15,18 we can assert that overall trend for higher sensitivity of ARPE-19 cells to contaminants was confirmed.
In summary, the results in this study indicate that the in vitro direct-contact cytotoxicity test was highly sensitive to hazardous contaminants even when they were tested at very low concentrations, and the dose-response curves obtained with this test allowed the cytotoxicity concentration of the tested substances to be determined. The in vitro direct contact cytotoxicity test may represent an additional tool to assess the safety of the PFCLs batch. Finally, the manufacturing processes of PFCLs should ensure the absence of impurities, and the safety evaluation should comprise whole range of validated methods including NMR spectroscopy, gas chromatography coupled with mass spectroscopy, in vitro cytotoxicity testing by direct contact, and animal model testing in the design phase, before the product can be considered for clinical use.