HOW TO ENSURE QUALITY
1
196 Much of the early work on the molecular biology of oncogenesis was done with the RNA tumour viruses of birds and rodents. Over the past few years it has been shown that the virally coded genes responsible for transformation are almost certainly derived from eukaryotic cellular genes. The cellular homologues of these genetic hitch-hikers picked up by RNA tumour viruses remain present in normal rodent and avian cells. Since no involvement of these viruses has been convincingly demonstrated in most human cancers, their relation to clinical oncology has remained dubious. However, in a recenweries of experiments, homology has been shown between a bladder carcinoma human oncogene and the cellular homologue of the Harvey murine sarcoma virus. 6 Similar homology was shown between a human lung carcinoma oncogene and the transforming gene of Kirsten murine sarcoma virus. 7 These results are of great biological interest, but have they any clinical implications? Little is known about the products of these cellular oncogenes, which must in turn be the triggers for malignancy. If we could find ways to control their function by drugs or other agents then we might discover novel systems for treating cancer. In a particularly elegant study, DNA fragments from a rat neuroblastoma were trans- fected into a mouse fibroblast line.8 After two passages the transfected mouse fibroblasts were injected into young mice and grew out as tumours. Antibodies appeared in the sera of these mice, which identified a polypeptide of about 185 000 daltons in molecular weight. It is thought that this protein was encoded by the transforming sequence and that it may mediate the transformation process in the rat neuroblastoma. Furthermore, this 185K molecule, like many viral trans- forming proteins, was a phosphoprotein. Indeed there is now evidence that the likely protein product of the human bladder oncogene derived from the EJ cell line is a 21K phospho- protein. Several groups are now applying the new tools of molecular biology to study human oncogene products and the prospects for logically developed cancer treatments have substantially improved. HOW TO ENSURE QUALITY THERE is a risk that the current enthusiasm for external quality-assessment schemes may distract attention from the other steps which laboratories must take to maintain the clinical value of their reports. The situation is aggravated by the confusion between quality control, quality assessment, and quality assurance.; Some helpful guidance comes from the book Good Manufacturing Practice,’ published for the pharmaceutical industry five years ago. Quality assurance, briefly, consists of a total system of organised working and testing designed to ensure the quality of the product: in the medical laboratory this will usually be a report. Quality control is a much more restricted term. It refers only to those tests and procedures which must be completed before a particular batch of results is issued. Proficiency testing is also part of quality assurance but is quite distinct from quality control. A complete quality assurance programme consists of five 5. Bishop JM. Enemies within: the genesis of retro virus oncogenes. Cell 1981; 23: 5-6. 6. Parada LF, Tabin CJ, Shih C, Weinberg RA. Human EJ bladder carcinoma oncogene is homologue of Harvey sarcoma virus ras gene. Nature 1982; 297: 474-78. 7. Rigby P. The oncogenic circle closes. Nature 1982; 297: 451-53. 8. Prodhy LC, Shih C, Cowing D, Finkelstein R, Weinberg RA. Identification of a phospho-protein specifically induced by the transforming DNA of rat neuro- blastomas. Cell 1982; 28: 865-71. 1. Editorial. Quality of performance in pathology laboratories. Lancet 1981; i:599. 2. Department of Health and Social Security. Guide to good pharmaceutical manufacturing practice. London: H. M. Stationery Office, 1977. main phases-which in practice overlap somewhat. A test must be designed to give reliable and useful results. It must be validated to ensure that it gives the same results as those of a reference method and that they are applicable to clinical problems. The third step, is to revalidate the test for use in the particular laboratory under actual working conditions. This involves checking arrangements for collecting, storing, and documenting specimens and for the recording, scrutiny, and issue of results. If these first three steps are gone through fully and perfectly it is arguable that no further measures are required so long as the procedures are unchanged. This is not a realistic suggestion, in an imperfect world. With every batch of tests quality-control systems must operate. These may include preparations of known potency or positive and negative controls. A positive control should, in general, be as weak as the weakest positive one wants to detect, and negative controls should be preparations containing those substances most likely to give invalid false- positive results. A good control system takes into account the stability and reliability (or lack of them) of the test used. It should be sufficient to reassure the laboratory scientist and the pathologist that results are valid but it must also be realistic. Even the best controlled Wassermann reaction will have its biological false-positives; no amount of quality control will make a non-specific screening test specific. There are less obvious opportunities for quality control. Laboratories must have procedures for verifying abnormal results before issue. An unexplained thrombocytopenia must be checked by blood-film examination and by inspection of the specimen for tiny clots; a positive HBsAg screen must be confirmed by neutralisation; and so on. If these procedures are built in to the organisation of a laboratory in such a way that reports cannot be issued until they have been completed, then they are part of quality control in its strict sense. The last part of quality assurance is monitoring of results. Here there is no accept-or-reject decision. Sensitivity, specificity, precision, and accuracy must be examined as time passes to ensure that the only trend is one of continuing improvement. Whitehead and Woodford3 have contrasted internal and external quality assessment. As internal programmes allow repeated testing of material they are particularly well adapted to measurement of precision (consistency of results). They are also useful where clinical material is scanty or unstable. A histopathologist can arrange to have a battery of slides reread under code even when he has insufficient tissue for national distribution. External quality assessment relies heavily on stable specimens. It is most effective at measuring the accuracy of a laboratory report. This may be judged against a reference method or simply against som kind of national (or other) consensus. Unfortunately, external quality assessment schemes rely on rather infrequent tests in each laboratory; thus it is never clear whether a bad result is caused by imprecision or inaccuracy, or by a combination of reasonable precision, reasonable accuracy, and bad luck. Finally the pathologist must talk to the clinicians. Did the report arrive in time? Did it arrive at all? Was the report legible and comprehensible? Did it help? A wholehearted pursuit of quality offers the medical laboratory scientist and the pathologist an opportunity to raise their complementary skills for the benefit of patients. 3. Whitehead TP, Woodford FP. External quality assessment of clinical laboratories in the United Kingdom. J Clin Pathol 1981; 34: 947-57
Transcript of HOW TO ENSURE QUALITY
196
Much of the early work on the molecular biology ofoncogenesis was done with the RNA tumour viruses of birdsand rodents. Over the past few years it has been shown thatthe virally coded genes responsible for transformation arealmost certainly derived from eukaryotic cellular genes. Thecellular homologues of these genetic hitch-hikers picked upby RNA tumour viruses remain present in normal rodent andavian cells. Since no involvement of these viruses has been
convincingly demonstrated in most human cancers, theirrelation to clinical oncology has remained dubious. However,in a recenweries of experiments, homology has been shownbetween a bladder carcinoma human oncogene and thecellular homologue of the Harvey murine sarcoma virus. 6Similar homology was shown between a human lungcarcinoma oncogene and the transforming gene of Kirstenmurine sarcoma virus. 7
These results are of great biological interest, but have theyany clinical implications? Little is known about the productsof these cellular oncogenes, which must in turn be the triggersfor malignancy. If we could find ways to control theirfunction by drugs or other agents then we might discovernovel systems for treating cancer. In a particularly elegantstudy, DNA fragments from a rat neuroblastoma were trans-fected into a mouse fibroblast line.8 After two passages thetransfected mouse fibroblasts were injected into young miceand grew out as tumours. Antibodies appeared in the sera ofthese mice, which identified a polypeptide of about 185 000daltons in molecular weight. It is thought that this proteinwas encoded by the transforming sequence and that it maymediate the transformation process in the rat neuroblastoma.
Furthermore, this 185K molecule, like many viral trans-forming proteins, was a phosphoprotein. Indeed there is nowevidence that the likely protein product of the human bladderoncogene derived from the EJ cell line is a 21K phospho-protein. Several groups are now applying the new tools ofmolecular biology to study human oncogene products and theprospects for logically developed cancer treatments havesubstantially improved.
HOW TO ENSURE QUALITY
THERE is a risk that the current enthusiasm for external
quality-assessment schemes may distract attention from theother steps which laboratories must take to maintain theclinical value of their reports. The situation is aggravated bythe confusion between quality control, quality assessment,and quality assurance.;Some helpful guidance comes from the book Good
Manufacturing Practice,’ published for the pharmaceuticalindustry five years ago. Quality assurance, briefly, consists ofa total system of organised working and testing designed toensure the quality of the product: in the medical laboratorythis will usually be a report. Quality control is a much morerestricted term. It refers only to those tests and procedureswhich must be completed before a particular batch of resultsis issued. Proficiency testing is also part of quality assurancebut is quite distinct from quality control.A complete quality assurance programme consists of five
5. Bishop JM. Enemies within: the genesis of retro virus oncogenes. Cell 1981; 23: 5-6.6. Parada LF, Tabin CJ, Shih C, Weinberg RA. Human EJ bladder carcinoma oncogene
is homologue of Harvey sarcoma virus ras gene. Nature 1982; 297: 474-78.7. Rigby P. The oncogenic circle closes. Nature 1982; 297: 451-53.8. Prodhy LC, Shih C, Cowing D, Finkelstein R, Weinberg RA. Identification of a
phospho-protein specifically induced by the transforming DNA of rat neuro-blastomas. Cell 1982; 28: 865-71.
1. Editorial. Quality of performance in pathology laboratories. Lancet 1981; i:599.2. Department of Health and Social Security. Guide to good pharmaceutical
manufacturing practice. London: H. M. Stationery Office, 1977.
main phases-which in practice overlap somewhat. A testmust be designed to give reliable and useful results. It must bevalidated to ensure that it gives the same results as those of areference method and that they are applicable to clinical
problems. The third step, is to revalidate the test for use in theparticular laboratory under actual working conditions. Thisinvolves checking arrangements for collecting, storing, anddocumenting specimens and for the recording, scrutiny, andissue of results.
If these first three steps are gone through fully and perfectlyit is arguable that no further measures are required so long asthe procedures are unchanged. This is not a realistic
suggestion, in an imperfect world.With every batch of tests quality-control systems must
operate. These may include preparations of known potencyor positive and negative controls. A positive control should,in general, be as weak as the weakest positive one wants todetect, and negative controls should be preparationscontaining those substances most likely to give invalid false-positive results. A good control system takes into account thestability and reliability (or lack of them) of the test used. Itshould be sufficient to reassure the laboratory scientist andthe pathologist that results are valid but it must also berealistic. Even the best controlled Wassermann reaction willhave its biological false-positives; no amount of qualitycontrol will make a non-specific screening test specific.There are less obvious opportunities for quality control.
Laboratories must have procedures for verifying abnormalresults before issue. An unexplained thrombocytopenia mustbe checked by blood-film examination and by inspection ofthe specimen for tiny clots; a positive HBsAg screen must beconfirmed by neutralisation; and so on. If these proceduresare built in to the organisation of a laboratory in such a waythat reports cannot be issued until they have been completed,then they are part of quality control in its strict sense.The last part of quality assurance is monitoring of results.
Here there is no accept-or-reject decision. Sensitivity,specificity, precision, and accuracy must be examined as timepasses to ensure that the only trend is one of continuingimprovement. Whitehead and Woodford3 have contrastedinternal and external quality assessment. As internal
programmes allow repeated testing of material they are
particularly well adapted to measurement of precision(consistency of results). They are also useful where clinicalmaterial is scanty or unstable. A histopathologist can arrangeto have a battery of slides reread under code even when he hasinsufficient tissue for national distribution.External quality assessment relies heavily on stable
specimens. It is most effective at measuring the accuracy of alaboratory report. This may be judged against a referencemethod or simply against som kind of national (or other)consensus. Unfortunately, external quality assessment
schemes rely on rather infrequent tests in each laboratory;thus it is never clear whether a bad result is caused byimprecision or inaccuracy, or by a combination of reasonableprecision, reasonable accuracy, and bad luck.
Finally the pathologist must talk to the clinicians. Did thereport arrive in time? Did it arrive at all? Was the reportlegible and comprehensible? Did it help? A wholeheartedpursuit of quality offers the medical laboratory scientist andthe pathologist an opportunity to raise their complementaryskills for the benefit of patients.
3. Whitehead TP, Woodford FP. External quality assessment of clinical laboratories inthe United Kingdom. J Clin Pathol 1981; 34: 947-57
