Leftover ethylene-diaminetetraacetic acid (EDTA)-anticoagulated blood samples from purebred dogs that had been submitted for routine diagnostic work to Laboklin GmbH& Co KG, Kissingen, Germany between December 2018 and October 2019 were used for this survey. Of a total of 206 samples, the majority (188) originated from Germany, with 18 samples from neighboring or other European countries (Czech Republic, Denmark, Finland, Italy, Luxembourg, Sweden, and Norway). The approval to use leftover samples for research was received from the Government IACUC in Bayern, Germany. Only leftover samples with of least 1 ml EDTA blood that was less than 1 week old and had been stored at 4–8°C prior to analysis were included. Samples from several dog breeds, classified based on the Fédération Cynologique Internationale (FCI¹), were selected based upon prior studies (11, 12, 27, 29, 36, 37) and our own early results of their varied blood group antigen expression. Samples from known related dogs or repeat samples from the same dog were excluded. Data on breed, gender, and age as well as geographic region were obtained. None of the samples showed any auto-agglutination in card and gel tests. All results were photographically captured.

Python v. 3.5.2 using Scipy² v. 0.17.0 and Numpy v. 1.11.0 were used for statistical analyses. Chi-square test was done using Scipy's Chi-square implementation. Results were considered significant when p < 0.05. Additionally, Cohen's Kappa Test in Scikit v. 0.17.0 was used for comparison of DEA 4 and DEA 5 Card and Gel Tests.

Utilizing the Strip with a monoclonal anti-DEA 1 alloantibody, we were able to identify dogs with no reactivity, and thus DEA 1–, as well as dogs that were strongly positive, with DEA 1 bands comparable to the positive control band. Of the 206 dogs, an equal proportion was either graded as DEA 1– (40.8%) or strongly DEA 1+ (40.3%) with a smaller proportion (18.9%) of weak to moderate DEA 1+ band reactions (Figure 1). Strong reactions (3+ and 4+) for DEA 1 were found significantly more often in Dalmatians (p = 0.0003) and Pugs (p = 0.0003), and weak/moderate (1+/2+) reactions were significantly more common in Labrador Retrievers (p = 0.001). All Pugs and Dalmatians except for one Dalmatian were DEA 1+, and more Labrador retrievers and Shih Tzus were DEA 1+ than were other breeds (Table 1).

In contrast, less than a quarter of Cane Corsos and Doberman Pinschers were DEA 1+, and all but one Boxer was DEA 1– (Table 1). These breeds belong to the FCI group 2. A significant dependency between breed and DEA 1 type was found for Cane Corso (p = 0.004), Dalmatian (p = 0.001), Doberman Pinscher (p = 0.001), Pug (p = 0.0002), and Boxer (p = 0.002).

When using the Card newly developed for DEA 4 typing according to manufacturer's instructions, 9.2% of 206 dogs showed an extremely weak and thus questionable 1+ agglutination reaction, and 67.0% produced only weakly positive (1+/2+) results. However, when adding an additional 40 μl of diluent, all 191 tested dogs produced ≥1+ and clearly visible agglutination reactions, supportive of a DEA 4+ blood type. When comparing the results between the two diluent concentrations, 72% of the reactions stayed the same, 18% strengthened, and 10% became weaker (never weakening by more than one degree, e.g., 2+ became 1+), but none of the positive agglutination reactions turned negative (Supplement Table 3). Based on these results, the manufacturer's instructions have been changed to indicate that if the results using the initial 40 μl are unclear, another 40 μl of diluent should be added.

To further evaluate the results obtained by the new DEA 4 typing card, we also applied the Gel using commercially available anti-DEA 4 alloantibodies (antiserum; same as for the Card). As all dogs tested were DEA 4+ with the Card and Gel, we had perfect agreement (Kappa = 1) between the tests. Of the 158 dogs tested, all reacted strongly positively (3+/4+) and, therefore, were considered to have blood type DEA 4+ (Figure 3). While all agglutination reactions observed by the Gel were 3+ to 4+, such strong reactions (3+ to 4+) using Cards were only seen in 23.8 and 15.7% using 40 and 80 μl of diluent, respectively. While the regression of DEA 4 agglutination strength with 40 μl and 80 μl in the 191 tested dogs showed a good correlation (r = 0.8999), there was no significant correlation of strength between the Gel and Card (r = 0.0754). Also, the degree of DEA 4 positivity did not appear to be associated with any particular breed.

When using the new DEA 5 typing card according to the manufacturer's instructions, 90.8% of 206 dogs did not show any RBC-agglutination reactions. Weak agglutination reactions (1+ or 2+) were seen in 19 dogs, and they were thus considered DEA 5+ (Table 1). The manufacturer states that the DEA 5 agglutination is “finer and less plentiful than that observed in a DEA 1 dog run on a DEA 1 test card” (DMS) and we also found the DEA 5+ card agglutination results to be consistently less intense than the DEA 4+ card results. Utilizing the Gel with the same polyclonal DEA 5 antibody used for the Card, 11.4% of the 158 dogs typed DEA 5+. With the Gel, six (33.3%) of the DEA 5+ dogs reacted only weakly/moderately positive, while the other 12 DEA 5+ dogs reacted strongly (3+/4+).

When comparing the Card and Gel methods, 97% of the DEA 5 results were concordant; in four of five discordant results, the Gel test indicated DEA 5+ (Table 2). The regression of DEA 5 Gel and Card strength in all 158 tested dogs showed a good correlation (r = 0.8329), and the two methods showed almost perfect agreement (Kappa = 0.831). For the Doberman Pinscher (p = 0.02) and Maltese (p = 0.03), the DEA 5+ blood type was significantly associated with the breed.

Using the Card, 9% of all 206 dogs and also 9% of the 158 dogs that were typed by both Card and Gel were DEA 5+, while 11% of those 158 dogs were DEA 5+ using the Gel. DEA 5+ dogs were found by the Card and Gel in Cane Corso, Dalmatian, Doberman, Lhasa Apso, and Maltese, and by the Gel also in one Boxer (Table 1).

Strong (3+/4+) agglutination reactions for Dal using a polyclonal anti-Dal antiserum in a Gel assay were observed in 183 of the Dal+ dogs, and a 2+ reaction was seen in one Shih Tzu that was on immunosuppressive therapy. In addition to the previously reported breeds with Dal– dogs, we found 22 Dal– dogs also in the breeds Cane Corso, Pug, and Mastiff (Table 1). We did not find any significant association between breed and Dal type, but the number of dogs per breed is small.

Utilization of the Gel and specific monoclonal antibodies against Kai 1 and Kai 2 resulted in either very strong agglutination reactions or completely negative test results (Figure 4). Overall, 96.6% of all dogs were Kai 1+/Kai 2– (Table 3). There were only five Lhasa Apsos and one Maltese that were Kai 1–/Kai 2+. Furthermore, only a single Maltese dog was found to be Kai 1–/Kai 2– in this survey (Table 1).

Overall, 40.8 % of all dogs in our survey were DEA 1–, which is very similar to recent surveys from USA, France, India, and Italy that also utilized the same immunochromatographic method and monoclonal anti-DEA 1 alloantibody (25, 27, 31, 37, 40). Comparison to prior studies is complicated, because these studies utilized polyclonal alloantibodies against DEA 1.1 and DEA 1.2 (7). However, recent studies clearly showed a close association between DEA 1 typing and DEA 1.1 typing results (19). In fact, the original monoclonal anti-DEA 1.1 alloantibodies used in DEA 1 typing kits were later renamed as DEA 1 by the manufacturers. The relationship between DEA 1 and DEA 1.2 is less clear, but it is likely that DEA 1.2 is a weaker DEA 1 antigen (20). No laboratories currently offer DEA 1.2 typing or any reagents for DEA 1.2 typing. Nevertheless, prior surveys using polyclonal anti-DEA 1.1 and anti-DEA 1.2 antisera also showed a predominance of DEA 1+ dogs (Table 4). A comparative overview of prior DEA 1 typing surveys is shown in Table 4.

Far more (71–90%) of the tested Cane Corsos, Doberman Pinschers, and Boxers were DEA 1–. Interestingly, these breeds belong to the FCI group 2, while other breeds studied were in other groups. Boxers and Cane Corso from various other European countries and South Africa were also mostly DEA 1– in prior surveys (details in Supplement Table 4). Thus, these breeds may be preferentially considered when recruiting blood donors to increase the DEA 1– donor pool, as suggested previously (25, 42)

In contrast, the Dalmatians and Pugs in our survey were all DEA 1+, except for one Dalmatian. Also, the Labrador Retriever and Shih Tzu breeds had mostly DEA 1+ dogs. The predominance of DEA 1+ dogs in these breeds was also observed in recent surveys from Switzerland, South Africa, Portugal, and India (details in Supplement Table 4). Finally, the Maltese breed had a near equal proportion of DEA 1+ and DEA 1– dogs, as previously reported from another survey in Italy (31). Thus, these DEA 1 typing surveys show similar proportions of DEA 1+ and DEA 1– dogs in each breed worldwide, which is not surprising considering their common ancestors and international breeding practices within breeds.

The strength of DEA 1 expression in DEA 1+ dogs is similar to a prior survey from the USA (27). Equal proportions are either strongly DEA 1+ or DEA 1–. Only a fifth of dogs show weak or moderate DEA 1 expression, an important observation for clinical practice, as the latter may lead to difficulties in interpretation and, potentially, in sensitizing dogs. Indeed, even blood from a weakly DEA 1+ donor given to a DEA 1– dog can elicit the sustained production of anti-DEA 1 alloantibodies (38), and sensitized dogs may develop an acute hemolytic transfusion reactions when again transfused (14).

All dogs in this survey were typed as DEA 4+ by the Gel and Card. Only a few prior typing surveys included DEA 4 typing, and these studies also revealed that almost all dogs are DEA 4+, as DEA 4– dogs are extremely rarely observed (7, 16, 27, 30, 37, 39). Thus, while DEA 4 is known as a blood group system, it should be considered a high-frequency antigen, as less than 1% of dogs are DEA 4– (14, 43).

The new DEA 4 typing card test was easy to perform, but some results were challenging to interpret. Agglutination reactions on the Card were weaker than those observed with the Control on the Card or with the Gel. The addition of another 40 μl diluent enhanced the weak agglutination reactions, and this phenomenon might be caused by a Prozone-like effect, though this was not further studied. The varying degrees of agglutination reactions observed with the Cards remain unexplained but are likely technical, as all 158 dogs tested produced strongly positive (3+ to 4+) results by the Gel method. Optimization of the Card to increase agglutination reactions would enhance the reliability of this assay. In the meantime, the manufacturer has independently adjusted the protocol based upon our observations.

Because the instructions are identical to DEA 1 typing cards, the DEA 4 cards can be readily added to the typing repertoire in clinical practice. The DEA 4 typing cards may be particularly useful for clinical diagnostics of incompatibility reactions in previously transfused DEA 4– dogs and for identifying DEA 4– donors. Indeed, early experimental studies (8) reported blood incompatibilities caused by DEA 4+ alloantibodies, and there is one clinical case report of an acute hemolytic transfusion reaction after the dog had been sensitized by a prior DEA 4+ transfusion (4). Thus, identifying DEA 4– donors could be very valuable when facing broad blood incompatibilities in previously transfused dogs potentially caused by DEA 4 sensitization.

In this survey from Germany, we found 9 and 11% DEA 5+ dogs with the Card and Gel, respectively. Similarly, a limited survey from the Netherlands identified 17% DEA 5+ (44), and other surveys from North America found 12–25% DEA 5+ dogs (14, 16). Although DEA 5 has been associated with blood type incompatibilities and reduced RBC lifespan experimentally (8), there are no clinical reports of any acute hemolytic transfusion reactions related to DEA 5. The new DEA 5 typing cards will be useful in the practical diagnostic examination of blood type incompatibilities and acute hemolytic transfusion reactions in previously transfused dogs.

Finally, the new commercially available DEA 4 and DEA 5 typing cards used in this survey produced similar but weaker agglutination results to the gel column method. Moreover, the agglutinations were weaker than seen with the commonly used DEA 1 cards. Nevertheless, there was a strong correlation between the two methods, with only five discordant DEA 5 typing results, supporting their use in clinical settings.

While original reports suggested that the Dal– blood type was exclusive to Dalmatians (11), recent studies from North America also found Dal– dogs in other breeds (27, 29, 36). Similarly, our survey from Germany found 10.7% Dal– dogs. This includes dogs from breeds with previously described Dal– dogs like Dalmatians, Dobermans, Shih Tzu, and Lhasa Apso (27, 29, 36), and also, for the first time, in Cane Corsos, Pugs, and a Mastiff. The prevalence of Dal– dogs in specific breeds is difficult to compare, because of the small numbers of dogs in each breed typed and reported geographical differences in the prevalence of Dal (29). Clearly, Dal typing is warranted for those breeds as well as for additional breeds, as Dal type blood incompatibility could potentially lead to an acute hemolytic transfusion reaction (29). Thus, it would be beneficial to identify Dal– donors for use in Dal– patients. Unfortunately, at this point, there is no commercial Dal typing kit available.

While the Gel uniformly showed strong (3+/4+) agglutination reactions with Dal+ blood, there was one dog that was receiving long-term treatment with immunosuppressives that had only a 2+ reaction. It may be possible that immunosuppression reduces the expression of the Dal antigen or the binding with the antibody. Differences in antigen expression have been shown in humans with illnesses (45) and were suspected in FeLV+ cats (46).

Utilizing the Gel, the agglutination reactions were very strong for Kai 1 and Kai 2. In this survey, 96.6% of all dogs were Kai 1+/Kai 2–, and few dogs were Kai 1– and Kai 2+ or Kai 2–. Thus, our results for the Kai 1 /Kai 2 blood type pattern distribution were similar to a prior survey from North America, which found 94% Kai 1+/Kai 2–, 5% Kai 1–/Kai 2–, and 1% Kai 1–/Kai 2+ (27). In a recent survey using the conventional tube agglutination assay with 203 mostly Mastiff dogs from South Korea, the proportion of Kai 1+/Kai 2– was only 42%, with the proportion of Kai 1–/Kai 2+ being 37% and that of Kai 1–/Kai 2– being 20% (12). The large group of Kai 1–/Kai 2+ Mastiffs seen in Korea could not be substantiated here and in the survey from North America because of a lack of samples received from this breed. Noteworthy, however, is the absence of dogs that were Kai 1+ as well as Kai 2+. While the Kai 1 and Kai 2 antigens appear to be very different based on protein size (12), it cannot be excluded that these two antigens are genetically related.

Finally, several Lhasa Apsos were found to be Kai 1– /Kai 2+ in the United Kingdom (Watson et al., personal communication). Because the only other Kai 1–/Kai 2+ dog was a Maltese and another Maltese was Kai 1–/Kai 2–, it is possible that these two related breeds of the FCI Group 9 are unique regarding their Kai antigen expressions.

However, the clinical importance of the Kai 1 and Kai 2 blood groups needs to be further investigated, especially since the development of alloantibodies against Kai 1 and Kai 2 in previously transfused dogs has been reported (12).

The authors declare that they are associated with either the Laboklin GmbH ' Co KG, Kissingen, Germany or PennGen Laboratory, University of Pennsylvania, Philadelphia, PA, USA. Both offer blood typing services. This survey is part of AE's doctoral thesis at the University of Zürich, Switzerland, with UG being her mentor. Laboklin provided the workplace and doctoral fellowship of AE and was involved in the study; the study received funding from Laboklin and was executed at Laboklin GmbH ' Co KG, Kissingen and Radolfzell, Germany, but was performed independently from Laboklin's routine laboratory work.