Abstract
An interview study was carried out with the aim of clarifying the reasons for the limited use of phenotypic characterization of genetically-modified mice (GMM) and identifying issues hindering its implementation. A total of 15 users of GMM participated in semi-structured face-to-face interviews, which were audio-taped and transcribed. The results were extracted using content analysis by theme. The investigation confirmed that few animals were systematically phenotyped and an observational approach was found to be widespread. The primary interest of the interviewees was phenotyping for impaired animal welfare. The concept of phenotyping was widely understood and perceived as a scientific advantage. The comprehensiveness of the protocols and the resources required for phenotyping were seen as problematic. All participants addressed this issue, be it regarding lack of time, money or expertise. Also, among the negative statements were worries about the capability of the available protocols to produce the information needed by the individual scientist. Phenotyping was predicted to become much more widespread in the future and its success was expected to depend on the development of reliable, fast and inexpensive methods. The study identified different aims of phenotyping and the suitability of the published protocols for these purposes was discussed. The contradiction between the limited use of characterization and its advantages was also discussed and proposals for the improvement of future phenotyping strategies are formulated.
Generation of genetically-modified mice (GMM) is not an exact science and in order to identify whether a genetic modification has led to an altered phenotype, it is necessary to test the animal and compare the results with a non-mutated mouse of the same strain. Testing GMM for new traits has been named phenotypic characterization or phenotyping and its benefits have been addressed by several authors.
1–5
Phenotypic characterization is necessary to ensure proper design of experiments. Only traits which have been identified can be taken into account when designing an experiment, therefore situations in which phenotyping is lacking increase the risk of obtaining compromised experimental results.
6
The risk of animal welfare impairment specifically related to GMM was already discussed at the beginning of the era of genetic modification and is still a topic of much interest.
7–12
An inventory by Thon
Jegstrup
Methods
Sampling
The study was explorative and a strategy of purposive sampling of participants was used in order to obtain maximum variation between participants. In this way we intended to identify the widest range of accounts rather than that the account should be numerically representative. This excludes quantitative analysis and firm conclusions but has the advantage of establishing the broadest possible testimony. 16 The following parameters, which were judged to influence the variety of answers, were taken into consideration in order to obtain this goal: size of institution employing the participant, public or private enterprise, age of participant, user/generator of GMM, nationality, previous interest in phenotyping and educational background (Table 1). A total number of 15 users of GMM were included. Fourteen participants were solely using GMM and one was both generating and using them. Out of these, three participants were responsible for animal facilities keeping genetically-modified animals at universities and other organizations and three were responsible for GMM in commercial companies. The remainder of the sample were scientists using GMM for experiments. Eleven participants were Danish citizens, and in order to obtain a wider perspective one participant each of Swedish, Finnish, British and Dutch origin were included. All respondents were promised anonymity when publishing the results.
Characteristics of participants in the survey
*Age of participant (1) <40, (2) 40–60, (3) >60 years
The investigation was based on a qualitative, cross-sectional, descriptive, social study by means of semi-structured, face-to-face interviews. The interviews were conducted in Danish with Danish citizens and in English with other nationalities and took place at the beginning of 2005. The interviewer had a DVM background and knowledge of GMM. She had previous training in communication, as well as experience in performing qualitative methods including interviewing and reporting in writing. Interviews lasted 40–60 min, were audio-taped and subsequently transcribed by the first author. Semi-structured interviews were used in order to encourage the interviewee to discuss phenotypic characterization and his/her knowledge about and attitude towards the topic.
An interview guide was used in order to cover all relevant topics (Table 2). The interview guide was thematically divided into three categories, each of which related to a particular theme. The topics/questions of the interview guide were based on the results of a literature study, 14 previous investigations 13 and discussions in the context of the project ‘Living with biotechnology – assessing the welfare of transgenic laboratory animals’ at the Faculty of Life Science, University of Copenhagen. 17 The first category of questions covered whether and, if yes, how the interviewee performed phenotypic characterization. The aim was to identify the current level of phenotypic characterization taking place at the particular animal unit. The second set of questions covered the actual theoretical knowledge about phenotyping. The third category dealt with the attitude towards characterization, in order to detect crucial factors in decisions made in the phenotyping process and to identify potential barriers. The questions were open and the interviewer attempted not to take a leading position but gently directed the conversation through the topics. Probing questions were used and in case new statements or topics emerged, they were included in the interview.
Interview guide
Two test interviews were conducted in advance, to allow improvement of the interview guide. These two respondents were responsible for a commercial facility and a scientist conducting experiments with GMM, respectively. The respondents were informed beforehand about the interview being a test interview, and that it would be decided afterwards whether or not to include the findings in the analysis. The interview guide originally included a poster on phenotyping meant as an inspiration for discussion, but it was taken away after the test interviews as it seemed to confuse the participants rather than promote the interview. The questions of the interview guide generated substantial information in the test interviews and no further changes were made. The outcome of the test interviews was validated, judged as satisfactory and included in the sample.
Data analysis
The results were extracted from the data using content analysis by theme. Individual transcripts were read several times, starting with a general reading to obtain an overall perspective of the topics, followed by a detailed identification of phrases and sentences identified by constant comparison during which each statement was checked and compared with the rest of the data to establish the analytical categories, which were hand-coded and indexed. In this way the coding frame was developed inductively as the data analysis progressed. The categories were further refined and reduced in number by thorough investigation of the data. Since the analysis of qualitative research data can require considerable interpretation by the investigator the coding of the transcripts and categories was double-checked to secure validity by a person with a relevant scientific background who was not involved in the investigation. Subsequently, the coding of the transcripts and categories were discussed and accepted.
Results
During the process it became clear that 15 interviews sufficed, as topics started to be repetitive after 10–12 interviews, and gradually no new topics emerged.
Characterization in practice
One of the findings emerging from the interviews was that phenotyping according to a written protocol was not frequently performed. Several respondents reported that they did not carry out a systematic characterization at all. The only systematic phenotyping was reported by a pharmaceutical company, which informed that all new GMM were tested for neurological deviations according to a protocol, which they had developed themselves and by a scientist at a university who had developed a basic welfare protocol in a joint effort with the Central Animal Unit of her institute. The latter had established humane endpoints based on the welfare assessment. One participant not phenotyping her animals knew of a colleague in the same institution who performed characterization.
We learned that a very commonly applied phenotyping approach was observational, based on the expertise of the animal technicians. The approach was not standardized or based on protocols. The animals were typically observed once per day when they were handled for other purposes and any deviation from normal was noticed and reported to the user. He or she was often the one to decide what to do following the discovery of a deviating phenotype. This approach was mainly used to ensure animal welfare but there was also a focus on new traits. Some used forms to register the observed deviations, others reported orally. In case of oral reporting it was impossible to compare recent and historical findings. The interviewees expressed great satisfaction with the skills of the animal technicians in this respect and felt that their need for characterization was met by this approach. Also, they expressed the point of view that observation by animal technicians was an economically realistic approach. One scientist was convinced that observation carried out by an animal technician, especially trained for this purpose was superior to a systematic protocol-based approach. The argument for this was that no matter how comprehensive a protocol, one can never be sure to detect everything. It was suggested that, instead of phenotyping every new strain according to a protocol, characterization should be carried out by a visual comparison to a control mouse during the daily routines and to proceed to a systematic characterization protocol only if deviations were spotted. This approach was seen as rather demanding regarding resource use, while carrying out a systematic characterization of every new strain was considered prohibitive in an economical sense.
The interviewees expressed a primary interest in characterizing animals for compromised welfare. Secondly, interest was focusing on single traits that were directly correlated to their specific scientific area. When tests were actually used for phenotyping the choice was guided by the scientific hypotheses or what was referred to as ‘common sense’. Some participants expressed an interest in basic characterization but a broad comprehensive characterization was not desired by anybody. However, the fact that commercially available animals were often characterized relatively well by the vendor was perceived as an advantage.
Knowledge about characterization
The principles of phenotyping according to a protocol were not widely known and few researchers had any experience with the practical procedures. Several participants were not aware that protocols/procedures for characterization were available or they knew that these existed but had no knowledge about their content. Some of the participants had read a protocol or they possessed knowledge about systematic phenotyping from other sources such as seminars. However, everybody expressed an understanding of the phenomenon of phenotypic characterization and had given it some thought indicating that they were aware that this was an issue but had taken no action to investigate this further. None of the participants were aware that phenotyping was commercially available.
Several participants knew where to find the results of phenotyping, but had never seen a document containing these results except for the two scientists who developed their own procedures and the single respondent buying commercial animals characterized by the vendor.
Attitude towards characterization
A positive attitude towards characterization and an understanding of the gains that it could provide were expressed in many interviews. Despite the fact that few users of GMM performed a systematic protocol-based characterization and despite their contentment with observational results, the general opinion of the participants was that they could benefit scientifically from systematic phenotyping. Welfare was considered most important for testing. Also gaining knowledge about new and perhaps unexpected traits was seen as an advantage. The participant who developed her own, mainly welfare-related procedure, expressed great satisfaction that she had invested time and resources in characterization before starting experiments as it allowed her to use the animals in an optimal way during the studies. However, far from all testimony regarding phenotyping was of a positive nature. One participant totally disagreed that characterization should be based on the static approach that a protocol represents. Instead, phenotyping an animal should be combining information developed by daily observation with the results of published experiments. It should be a continuously ongoing process and relevant information should constantly be added to it. Also it was stated that characterization of a strain might create a dilemma since you could find that your model was not the best for your particular experiment and you would then have to decide whether you wanted to use it anyway or spend time/resources on finding a better-suited model.
The remainder of the negative statements was related to the instruments (protocols) for phenotyping but not to the concept. Some participants believed that systematic phenotypic characterization would only very rarely lead to useful findings relevant to their particular research and that the investment would be large compared with the benefits.
All interviewees addressed the issue of resource use – be it time, money or expertise. The comprehensiveness of some protocols was seen as a huge problem and the participants expressed the opinion that they needed three to five tests, which would give them a basic knowledge of their GMM. The financial problem was seen as a central obstacle and participants funded by grants stated that they were in a situation of having to produce convincing scientific results within a limited time frame in order to obtain the next grant. Characterization was not seen as a convincing result by the funding bodies and often made the scientists choose to not spend resources on this. Several participants also addressed the issue of time. A comprehensive characterization was seen as too time-consuming and detrimental to publication of up-to-date results.
The participants expected authorities to increase demands regarding phenotypic information about GMM. Further regulation/legislation was mainly seen as an unnecessary limitation to scientific progress. The future of phenotyping was expected to depend on whether it could be proven to be advantageous to the scientist. It was mentioned that scientific journals would play a role in this by requesting phenotypic results of GMM before acceptance of a paper. It was stated that an unfortunate tendency was seen that the description of GMM in publications had become less detailed and exact than only a few years ago. The need for expertise to perform several of the phenotyping tests was seen as a problem. The lack of veterinary histopathologists was mentioned and so were the difficulties in properly performing standardized behavioural tests. Furthermore, possessing the know-how to choose the relevant parameters for characterization was seen as an obstacle. Some believed that the future of phenotyping was dependent on the development of new methods. In case simple, dependable and inexpensive tests would be developed, phenotyping would have a great future. More non-invasive and gentle methods should be developed. Finally, the opinion was expressed that such large numbers of GMM would be generated that it would be impossible to systematically characterize them all.
Discussion
The interviewees of this study all contributed with substantial and relevant information enabling us to establish a wide perspective on phenotyping as aimed for. We have demonstrated that few of the GMM in this investigation were systematically characterized. This corresponds well with the findings of a literature search by Jegstrup
Phenotyping for impinged animal welfare was important to the participants of this investigation and the question is if welfare assessment should be an individual exercise or if it would be sufficient to test for new traits and base the welfare assessment on this. Since the phenotype of any genetic modification is a complex interaction between gene allele and genetic background, one can never be sure that an expected phenotypic change has taken place and the phenotyping strategy should be such to optimize the chance of detecting both expected and unexpected effects of the mutation. In case this broad approach is taken, the welfare assessment of the animal is probably covered, but if focus is on one trait only, as often seen in this survey, a regular welfare assessment should be carried out as well.
Executing a welfare assessment as the one suggested by the British National Centre for Replacement, Refinement and Reduction of Animals in Research (NC3Rs) 12 would provide the user of a GMM with the necessary information for relevant authorities. At the same time it would secure information about special problems of a particular strain, thereby allowing the animal technician to take special measures to improve its welfare, e.g. special diet, special housing or medication. Ultimately, this will improve animal welfare, enable better study design and prevent the loss of valuable strains. Also, humane endpoints can be based on welfare assessment.
Some interviewees were not aware of the existing protocols, which led to the development of own methods and more often to lack of characterization. When individual laboratories adhere to different procedures it becomes impossible to compare the results obtained. Both incomparable results and lack of phenotyping are non-beneficial to animals and scientists, and more information about phenotyping according to standardized protocols is needed.
Concern about the comprehensiveness of the published protocols was stated quite clearly, and spending the resources needed for systematic phenotyping was perceived as highly problematic. The issue of comprehensiveness of characterization has been addressed by Champy
In summary, this study has provided a solid base for improvement of phenotyping. We believe that an effort within (i) optimal design and use of protocols; (ii) development of valid high-throughput methods; (iii) cost–benefit analysis would be of benefit. In this way more effective phenotyping strategies could be developed, safeguarding both scientific and animal welfare demands.
Footnotes
ACKNOWLEDGEMENTS
The first author wishes to thank Novo Nordisk, Denmark, for their generous sponsorship of her PhD study of which this investigation was a part. Also, she would like to express her gratitude for allowing her to use their facilities during the writing phase of this paper.