Tears collected with an absorbent material and extracted with centrifugation yielded a TPC of 5.2–14.6 mg/mL in dogs and 6.2–20.6 mg/mL in cats. These values are likely accurate estimates of the ‘true’ TPC for samples in which the tear volume absorbed was around 20–25 μL (i.e. median VA in vivo), but may represent an over-estimation of TPC for samples with low VA and, to a lesser degree, those with high VA. Indeed, our in vitro data showed a clear ‘concentrating effect’ from both absorbent materials when VA was low (10 μL), which was noticeable across the entire range of protein levels tested and more pronounced with Schirmer strips than PVA sponges. Despite centrifugation, our group has previously shown that a relatively large portion of fluid was retained by ophthalmic sponges when the volume of tears absorbed was low , especially for cellulose-based sponges. That explains the greater concentrating effect noted in the present study for Schirmer strips, which are hydrophilic materials manufactured from cellulose filter paper .
Regardless, the level of proteins reported herein is superior to the one described in tear fluid collected with capillary tubes in dogs (2.6 mg/mL)  and cats (10.4 mg/mL) . Consistent with older reports in human subjects, the higher TPC in absorbent materials such as Schirmer strips vs. capillary tubes is presumably related to an increased in serum-derived proteins due to conjunctival irritation [7, 17]. However, these findings are not consistent across the scientific literature as recent studies have described an equal  or inferior  amount of proteins in Schirmer-collected tears when compared to capillary tubes. Thus, it is likely inaccurate to compare the findings of various studies based on the tear collection method alone. Rather, one should also consider how the tears were extracted from the absorbent material - i.e. centrifugal force , elution with a solvent [8, 15], or a combination of both [18, 19] - as well as the method used by the investigator to quantify the protein content.
In the present study, the assay used to quantify TPC relies on infrared spectrometry that measures the amide bond absorbance across all proteins and peptides contained in the sample. This method improves the accuracy of the assay as compared to colorimetric methods . Although extensively used for TPC in tear fluid because of their simplicity and sensitivity [3, 8, 16], colorimetric methods such as Lowry and Bradford have drawbacks that can result in under- or over-estimation of TPC in tears . For instance, the Bradford assay highly depends on the specific protein composition in a sample , and this becomes a considerable disadvantage when evaluating TPC in a complex protein mixture such as tear fluid. Further, since serum albumin is not a major component of tear proteins [23, 24], using BSA as a standard likely affects the TPC obtained by the colorimetric assay; in fact, TPC in tear fluid significantly differs whether BSA or IgG is used as the standard for the method .
A significant negative correlation was noted between tear flow rate and TPC in dogs, consistent with findings by Fullard and colleagues . In another canine study, individuals with epiphora consistently had low tear fluid protein levels while subjects with dry eye had a high protein concentration . In contrast, there was no correlation between tear flow rate and TPC in our feline subjects, and a recent publication in human patients showed a positive correlation between TPC and the Schirmer readings . The reason for these discrepancies remains unclear and warrants further investigation. Further, body weight and age were not correlated with TPC in either species, despite the reported positive correlation between body weight and tear volume absorbed in dogs . This finding, as well as potential impact of breed on TPC, should be verified in a larger population size.
Importantly, our study found that the repeatability of protein quantification in tears was greater with Schirmer strips when compared to PVA sponges, as determined by significantly lower inter-session CV% in both species. This could be explained by the lower variability in tear volume absorbed by Schirmer strips. Indeed, the mm- marks of Schirmer strips can be used to standardize the volume of tears absorbed at each collection. In the present study, the Schirmer strips were left in the conjunctival fornices of dogs and cats until the 20-mm mark was wetted, thus reducing the variability in VA between sessions and therefore improving the repeatability of TPC measurement. On the other hand, standardizing the duration of PVA-collection to 60 s did not reduce the variability in VA obtained between sessions; for instance, the VA in one representative dog was 20 μL at the initial sampling, and 43 μL when tear collection was repeated 10 min later. Although not evaluated in the present study, the extraction of tears from Schirmer strips could be optimized by ‘washing’ the strip with a solvent to help retrieve any small amount of protein that remained on the absorbent material post-centrifugation .
The present study was limited to describing total protein content, i.e. ‘gross’ protein quantification in the tear fluid. Further studies could investigate which specific tear proteins (e.g. albumin, lactoferrin) are most affected by changes in tear flow rate, or by the adsorptive properties of the Schirmer strips and PVA sponges used for tear collection. A recent study in humans determined that proteins are preferentially retained by Schirmer strips based on physicochemical factors such as molecular weight and surface charge .
Nevertheless, results of the present study are valuable given the potential applications of total protein quantification in veterinary medicine. In a clinical setting, TPC can be used to better characterize various ocular surface diseases and assess their response to therapy, as exemplified by keratoconjunctivitis sicca in dogs  and corneal sequestrum in cats . In a research setting, TPC can be used to standardize the amount of tear sample to be used for assessing inflammatory mediators , in-depth proteomics  and others.