Vol. 89 • No. 5
hand, eosinophil granules appear as
large rounded or oval granules which
consist of electron dense homogeneous
material and occasional crystalloids.
However, in our experience, it seems
difficult at times to identify hybrid granulocytes because immature or defective
eosinophil granules resembled immature or defective basophil granules. Although the magnification of the electron
micrograph illustrated as a hybrid granulocyte by Weil and Hrisinko 8 seems insufficient for exact evaluation, the cell
seems somewhat different from typical
hybrid granulocytes; a granule indicated
as a basophil granule appears somewhat
lacking the features characteristic of basophil granules. It seems to represent an
immature or defective eosinophil granule rather than a basophil granule. In
this connection, it should be noted that
eosinophils possessing immature or defective eosinophil granules are frequently present in CML. We show a hybrid granulocyte possessing both typical
basophil granules and eosinophil granules (Fig. 1). In our experience, crystalloids were rarely seen in hybrid granulocytes.
For exact identification of hybrid
granulocytes, electron microscopic ruthenium red staining according to the
method of Kobayasi and Asboe-Hansen2
appears suitable. For this staining, we
have fixed and dehydrated aspirated
703
CORRESPONDENCE AND CORRECTIONS
bone marrow particles in 100% ethanol
or methanol, and then embedded directly in Epok 812 (Oken, Inc.). According to this method, hybrid granulocytes
simultaneously show ruthenium red
positive granules and negative granules.5
In summary, the presence of hybrid
granulocytes suggests the following possibilities: (1) Basophils and eosinophils
are derived from a common precursor
cell. (2) The differentiation pathways for
either basophils or eosinophils are more
closely related to each other than we
have believed. (3) Abnormal or defective
immature basophils may retrieve the
ability to differentiate into eosinophils
during the process of maturation possibly due to deranged intrinsic factors
(e.g., abnormal gene activation or derepression) or environmental factors (e.g.,
various biological stimulating factors)
or, conversely, abnormal or defective
eosinophils may retrieve the ability to
differentiate into basophils.
NOBUO TAKEMORI,
NAGAHITO SAITO,
M.D.
M.D.
NORIKO TACHIBANA,
M.D.
NAOYUKI HAYASHISHITA,
TAMOTSU MIYAZAKI,
M.D.
M.D.
Third Department
of Internal Medicine,
Hokkaido University
School of Medicine,
Sapporo, Japan 060
References
1. Doan CA, Reinhart HL: The basophil
granulocyte, basophilcytosis, and myeloid leukemia, basophil and "mixed
granule" types: An experimental, clinical and pathological study, with the
report of a new syndrome. Am J Clin
Pathol 1941;11:1-33.
2. Kobayasi T, Asboe-Hansen G: Ruthenium red staining of ultrathin sections
of human mast-cell granules. J Microsc 1971;93:55-60.
3. Mlynek ML, Leder, LD: Zur Kenntnis
und zur Bedeutung hybrider Granulozyten bei chronischer myeloischer
Leukamie. Schweiz med Wschr
1985;115:1086-1091.
4. Parwaresch MR (ed): The human blood
basophil. Berlin, Springer-Verlag,
1976, pp 203-230.
5. Takemori N, Saito N, Musashi M, et al:
Light and electron microscopic study
of hybrid granulocytes in chronic myeloid leukemia. Igaku no Ayumi
1986;139:843-844.
6. Takemori N, Saito N, Tachibana N, et al:
Hybrid granulocytes in chronic myeloid leukemia: Light and electron microscopic study. J Clin Electron Microsc 1987;20:143-149.
7. Tanno Y, Bienenstock J, Richardson M,
LeeTDG, BefusAD, DenburgJA: Reciprocal regulation of human basophil
and eosinophil differentiation by separate T-cell-derived factors. Exp Hematol 1987;15:24-33.
8. Weil SC, Hrisinko MA: A hybrid eosinophilic-basophilic granulocyte in
chronic granulocytic leukemia. Am J
Clin Pathol 1987;87:66-70.
The Author's Reply*
Analytic Goals Are Targets, Not
Inflexible Criteria of Acceptability
To the Editor:—The comments of
Drs. Barnett and Skendzel on my proposal5 for the setting of analytic goals for
hematology tests from data on biologic
variation have common aspects, so I
shall answer these in this single response.
Dr. Barnett argues that the adoption
of my proposal would narrow "presently
accepted" goals. In fact, as recently
stated by Hackney and Cembrowski,"
there are no widely accepted estimates of
allowable error for hematology tests.
* Author's reply to letters of Barnett and
Skendzel that appeared in the April 1988
issue.
The goals advocated5 are indeed more
stringent than those previously suggested by Skendzel, Barnett and Piatt.15
In my opinion, and in the view of
others, 212 their proposals are interesting,
but the concept is flawed. The proposals
are based on the responses of physicians
to clinical vignettes and imprecision calculated from what physicians view as a
significant change. Since, in the clinical
situation, changes in serial results depend on preanalytic plus biologic plus
analytic variation, estimates of goals assuming that such changes are caused
solely by analytic imprecision are bound
to be inappropriately high. Moreover,
goals derived from the opinions of clinicians probably depend on the specific
question posed and on the previous individual experience of the clinician with
the laboratory services provided. I therefore support the view of Batsakis1 that
formulation of analytic goals by collation and quantitation of surveys (of clinicians) is not very productive and often
is misleading.
If present performance goals are
based, as alleged by Dr. Barnett, on the
state-of-the-art performance achieved
with the best equipment in wide use,
then these are inappropriate. The state
of the art generally is derived from inter-
704
laboratory quality assessment schemes;
unfortunately, the materials circulated
to participants may not behave exactly
as specimens from patients,8 and laboratories may analyze the specimens under
special conditions.14 Moreover, if the
state of the art achieved with the best
equipment is used to set goals, then goals
will become more and more stringent
with time as instrumentation improves.
Instrument manufacturers and laboratories alike will be attempting forever to
attain these ever-changing goals. This is
a waste of scarce resources. In situations
in which analytic performance meets or
surpasses objectively set analytic goals,
further improvement of instrumentation and methodology is not required,
and available efforts, skills, and funds
should be directed toward improvement
of methods of analysis of analytes for
which goals are not met.
Dr. Barnett suggests that clinicians
have no interest in biologic variation in
the healthy person. This unfortunately
has a ring of truth at the present time,
but, as clinical laboratory scientists, we
all must attempt to correct this deficiency through education, as has been
attempted in the tertiary care teaching
hospitals in which I have served.4 There
are estimates of biologic variation for
many analytes in the serum, blood and
urine of healthy subjects.6 The estimates
are remarkably consistent irrespective of
the time period over which studies were
performed, the country in which the
studies were done, and the analytic technique used. Since it has been shown that
the biologic variation in stable disease is
of the same order as in health, this large
data base should be applied in everyday
practice to set analytic goals, to decide
on the true utility of conventional population-based reference values,10 and
most importantly, to decide on what
constitutes a significant difference between serial results in an individual (the
critical difference). Dr. Skendzel thinks
that physicians have not changed their
attitudes on what constitutes a significant change as methods have improved
because they have found that investigation is not rewarding. I strongly favor his
alternative explanation that physicians
are unaware of the importance of biologic and analytic variation and that
the average critical difference is (for
P < 0.05) equal to 2.77 (imprecision2
A.J.C.P. • May 1988
CORRESPONDENCE AND CORRECTIONS
+ biologic variation2)1'2; it is on this particular topic that education is sorely required.
Dr. Barnett considers that analytic
goals can be used to set acceptability
limits for proficiency testing surveys. I
support this view and the contention of
Gilbert9 that any proficiency testing program that declares a result unacceptable
must, at some time, face the question of
defining acceptability. Recently it has
become realized that, as elegantly demonstrated by Erhmeyer and Laessig,3 acceptability limits have many deficiencies
irrespective of whether they are set using
the state-of-the-art (±2 SD or ±3 SD
limits) or fixed limits.
The real problem, as stated by Dr.
Barnett, is that, in the United States,
proficiency surveys are used as regulatory tools; the limits of acceptability set
by Dr. Boone and his colleagues at CDC
may be designed so that a very high proportion of laboratories are deemed acceptable. In contrast, in the United
Kingdom17 such surveys are designed to
facilitate interlaboratory transferability
of results, help the laboratory in the selection of appropriate instrumentation
and methods, and provide expert advice
on the solution of individual problems
and difficulties. This subject was discussed in detail at the 1987 Jackson
Conference of the College of American
Pathologists; it was suggested there that
biologic variation-based goals should be
used as limits of acceptability when the
majority of laboratories could meet
these but that the state of the art should
be used until such goals were met. Importantly, it was stated that objective
goals should be detailed to all participants in all communications of results.
The advocated goals should not be
used ubiquitously as limits for acceptable intralaboratory imprecision of all
analytes. Correct application would, in
fact, save costs. When analytic performance meets or surpasses goals, the
goals should be used as the criteria for
rejection or acceptance of runs in preference to the currently applied statistical
quality control rules.1316 The simple ±3
SD rule then can be used to set the
"warning" limits rather than "action"
limits, fewer runs will be rejected, and
significant resources saved. In contrast,
if analytic performance does not attain
the goals, the usual statistical rules
should be applied as at present; it is to
such analytes that resources should be
applied and efforts made to improve performance.
It is more difficult to quantitate the
cost benefits of applying the advocated
goals in patient care, and this topic definitely is worthy of detailed study. However, a number of points are relevant.
Since most tests done in hospital laboratories are used for monitoring patients
rather than for diagnosis, I believe that
attainment of the goals has advantages
in that the critical difference between serial results is dependent on imprecision
as well as biologic variation; large imprecision makes for large critical differences, and the physician will be unsure
whether any change seen is caused by
improvement or deterioration and may
prolong expensive hospital stays until
the position has become clear. In other
specific clinical situations, imprecision
also is critical: for example, in deciding
whether the concentration of a tumor
marker is truly increasing so that prompt
therapy or surgery may be undertaken to
eliminate significantly more expensive
later care of the patient with extensive
disease. Similarly, we previously have
discussed the ramifications on costs of
different levels of imprecision on analyte
levels that decrease with time.7
I therefore believe that the advocated
goals are worthy targets to strive to attain, provided that they are used as the
ideals and not as totally inflexible criteria of acceptability as they appear to be
viewed by Drs. Barnett and Skendzel.
G. FRASER, PH.D., F.A.A.C.B.
Top Grade Biochemist/Senior Lecturer,
Ninewells Hospital and Medical School,
Dundee, DD1 9SY,
Scotland
CALLUM
References
1. Batsakis JG: Analytical goals and the College of American Pathologists. Am J
Clin Pathol 1982;78:678-680.
2. Boerma GJM: Questionnaire highly interesting: Conclusions make little
sense. Am J Clin Pathol 1986,85:392.
3. Erhmeyer SS, Laessig RH: An assessment
of the use of fixed limits to characterize interlaboratory performance by
proficiency testing. Clin Chem
1987;33:1901-1902.
4. Fraser CG: Interpretation of clinical
chemistry laboratory data. Oxford,
Blackwell Scientific, 1986.
Vol. 89 • No. 5
5. Fraser CG: Special Report. Desirable
standards for hematology tests: A proposal.
6. Fraser CG: The application of theoretical
goals in proficiency testing. Arch
Pathol Lab Med, in press.
7. Fraser CG, Browning MCK: Deciding
the optimum interval between specimen collections: theory and nomograms. Clin Chem 1987;33:14361438.
8. Fraser CG, Peake MJ: Problems associated with clinical chemistry quality
control materials. CRC Crit Rev Clin
LabSci 1980;12:59-86.
9. Gilbert RK: CAP Interlaboratory survey
data and analytic goals. In: Elvitch
FR, ed. Analytical goals in clinical
chemistry. CAP Aspen Conference,
CORRESPONDENCE AND CORRECTIONS
10.
11.
12.
13.
14.
1976. Skokie, Illinois: CAP, 1977:6373.
Harris EK: Statistical aspects of reference
values in clinical pathology. Prog Clin
Pathol 1981;8:45-66.
Hackney JR, Cembrowski GS: Need for
improved instrument and kit evaluations. Am J Clin Pathol 1986,85:391393.
Ross JW: Goals for allowable analytic
error better based on medical usefulness criteria. Am J Clin Pathol
1986;85:391-392.
Ross JW, Fraser MD: Clinical laboratory
precision: The state of the art and
medical usefulness based internal
quality control. Am J Clin Pathol
1982;78:578-586.
Rowan RM, Laker MF: Alberti K.GMM:
705
The implications of assaying external
quality control sera under 'special
conditions'. Ann Clin Biochem
1984;21:64-68.
15. Skendzel LP, Barnett RN, Piatt R: Medically useful criteria for analytic performance of laboratory tests. Am J Clin
Pathol 1985;83:200-205.
16. Stamm D: A new concept for quality
control of clinical laboratory investigations in the light of clinical requirements and based upon reference
values. J Clin Chem Clin Biochem
1982;20:817-824.
17. Whitehead TP, Woodford FP: External
quality assessment of clinical laboratories in the United Kingdom. J Clin
Pathol 1981;34:947-957.