Case 5 -
Immunohistology in Dermatopathology: Problem Areas
Mark Robert Wick, Univ of Virginia Health System, Charlottesville, VA
Click on slide thumbnail images for an enlarged view.
If you have any difficulties viewing these slides, email the webmaster.
Illustrative Case: A 20 year old woman presented with a 0.6 cm
flesh-colored nodule in the left paranasal skin, which had slowly enlarged over the previous six months.
Physical examination showed no other abnormalities, and the patient was otherwise well. The lesion was
removed by excisional biopsy. Pathologic evaluation showed a neoplasm comprising epithelioid and
bluntly-fusiform cells with moderate nuclear atypicality and mitotic activity; very focal ulceration was
evident but no pigmentation was apparent. There was no lesional growth involving the dermoepidermal
junction. Immunostains were obtained, showing reactivity for vimentin, S100 protein (with nuclear
labeling only), and NKI/C3. Melan-A, HMB45, desmin, muscle-specific actin, pankeratin, and epithelial
membrane antigen were absent. A diagnosis of melanoma was made and the patient was referred to the
presenter's institution for sentinel lymph node biopsy and enrollment in a melanoma vaccine trial
study. Because of her young age, a request was made by the local attending physician to repeat the
panel of immunostains. This was done, with reproduction of the original results except for a negative result for S100 protein (both
with a heteroantiserum and with a mixture of monoclonal antibodies). An additional stain for S100A6 was
positive, and the diagnosis was subsequently amended to that of cellular neurothekeoma.
Case 5 - Figure 4
An epithelioid and spindle-cell neoplasm is present in the dermis with variably-prominent nucleoli and mitotic activity.
Case 5 - Figure 5
An epithelioid and spindle-cell neoplasm is present in the dermis with variably-prominent nucleoli and mitotic activity.
Case 5 - Figure 6
Immunostains on the dermal tumor (labeled individually).
Case 5 - Figure 7
S100 immunostain done at the presenter's institution, showing no reactivity in cellular neurothekeoma. This result was obtained with both a heteroantiserum and a mixture of monoclonal antibodies to S100 protein.
Case 5 - Figure 8
Positive immunostain for S100A6 protein in cellular neurothekeoma.
Familiarity breeds contempt… or does it, really? At this point, diagnostic
immunohistochemistry (DIHC) has been part of the armamentarium of surgical pathology for over thirty
years. Accordingly, practitioners have had ample opportunity to learn about the gamut of potential
results which one may obtain in applying that technique to clinical material. And yet, less procedural
acquaintanceship than ever exists with the specific methodological elements that comprise DIHC, at the
level of physician-pathologists. That situation is nothing new and rather mundane, and it probably
reflects a generational shift away from the technical training that was emphasized in pathology
residencies through the 1980s. Especially with the currently-easy availability of automated
histochemistry, immunohistochemistry, and in-situ hybridization in clinical settings, laboratory
physicians do not necessarily concern themselves with procedural details if they do not wish to do so, or
unless and until some sort of cataclysmic and systematic error appears in their midst. Put another way,
if one has no "au fait," one lacks the knowledge to have "mépris."
What are the potential outcomes of this state of affairs? At a system-level, one would
hope that participation in external quality assurance programs should reveal recurring procedural flaws
that could negatively affect patient care; hence, the more of those programs one participates in the
better. Nevertheless, they are by no means a fail-safe protection against method-related mistakes that
can occur on a single-case level. Only by rigorous and ongoing attention to the sub rosa details of DIHC can one hope to minimize those problems, and they will
serve as the focus for the remainder of this discussion. If one considers the traditional tripartite
scheme of preanalytical, analytical, and postanalytical (interpretative) steps in the diagnosis of any
given case, potential problems can be attached to DIHC at each of those levels.
It is tempting to presume that advancements in laboratory technique have compensated for variations
in tissue fixation, but such a supposition would be tantamount to magical thinking. It has been, and
probably always will be true that tissues which are not placed in fixative for a prolonged period after
excision will demonstrate a notable deterioration in antigenicity. Moreover, the particular type of fixative—e.g.,
coagulating (Bouin's solution; ethanol; Carnoy's solution; B5 solution; Zenker's fixative),
non-coagulating/cross-linking (formalin; glutaraldehyde; paraformaldehyde), or neither (microwave
fixation)—has a potentially-major influence on the preservation of various protein molecules which are
the targets of reagent antibodies. One can usually control the variable of fixation reasonably well
inside a local or regional healthcare system, but when case material is regularly received from
laboratories in other locales it is often impossible to know how that tissue has been preserved and
processed. This can lead to insidious variation in immunostaining results that produces possible
technique-related errors in eventual interpretation. Hence, before concluding that a referring
laboratory has erred in producing a particular result in DIHC, it would be well to learn the details of
their procedures, including the nature and length of tissue fixation.
In the analytical phase of immunohistology, several factors can once more "hide" under
cover of automation in the laboratory, but exert a negative influence on the final staining product. One
may use any one of several wax-clearing agents in deparaffinizing tissue sections, including xylene,
toluene, AmericlearR, HistoclearR, and other solutions. Some of them may produce a
subtle alteration is selected tissue antigens that either enhances ultimate immunoreactivity or decreases
it. Antigen "retrieval" steps can employ proteolytic enzymes, urease, wetting agents, or a heating step
(in a microwave oven, pressure-cooker, or steamer), with the latter including immersion of tissue
sections in buffer solutions. Those, in turn, can vary substantially in chemical composition, ionic
strength, and pH, all of which must be controlled within relatively narrow boundaries. Heat-induced
epitope retrieval (HIER) augments the intensity of immunolabeling for some markers and decreases it for
others, vis-à-vis DIHC procedures that omit an unmasking step. But, what, exactly, is being "undone" by
proteolysis or HIER? To this day, the answer to that question is still vague. Several hypotheses have
been advanced to account for epitope-unmasking. These include the breakage of fixation-induced coupling
of "irrelevant" but sterically-interfering large proteins to peptide epitopes; the abrogation of
electrostatic, van der Waal-like charges between epitopes and Fab fragments of reagent antibodies;
dissolution of cage-like calcium complexes around epitope sequences; and a reversal of Mannich reactions
between proteins. The latter are organic reactions featuring the amino- alkylation of acidic protons,
placed next to carbonyl groups during formaldehyde fixation.
Several potential immunodetection systems are likewise available for DIHC and they are associated
with variable levels of sensitivity, especially for antigens of low density. Some commonly available
possibilities include the peroxidase-antiperoxidase (PAP) procedure; the avidin-biotin-peroxidase complex
(ABC) method and its offshoots; the ABPAP technique (serially-combined PAP and ABC procedures); the
alkaline phosphatase-antialkaline phosphatase (APAAP) method; and the dextran-polymer-based
(EnvisionR) technique. The last-named method obviates the need for separate labeled secondary
antibodies from differing animals, and, at the same time, it greatly increases final immunostaining
After the development of chromogenic precipitate (of diaminobenzidine or aminoethylcarbazole) has
occurred in DIHC preparations, some laboratories employ chemical "enhancer" solutions (e.g., osmium tetroxide; copper sulfate; nickel chloride) to darken the final
staining product. Unfortunately, that eventuality may also intensify unwanted "background" reactivity
and lead to interpretative mishaps. Thus, it is germane to know whether enhancers have been used in
producing the slides in any given case.
At this point, I must say that all of the foregoing points apply regardless of which "platform" one
uses for DIHC. In other words, "old-fashioned" manual staining techniques are no more prone to such
pitfalls than any of the automated commercial systems that are presently employed in the clinical arena.
Automation does not abolish the commission of error—it merely
The interpretative phase of DIHC continues to be the most problematic. I will preface my
remarks in this section by saying straightaway that they will probably sound pompous (or at least crabby)
to many people. However, I have now been practicing surgical pathology and doing immunohistology "from
the ground up" for 30 years, and my experiences during that time comprise the grist of my comments.
One of the principal problems now extant in the clinical practice of pathology is that
people order immunostains with no clear plan attached to the process. Long lists of
biologically-disparate markers are often obtained, before considering what consequences possibly-positive
results will bring. I fear that this practice derives from two serious problems. The first is that
education in morphological diagnosis has been deemphasized in pathology residencies and fellowships.
This produces a needless and very-possibly misleading overreliance on various adjunctive technological
methods. Secondly, pathologists have allowed themselves to be convinced that clinicians, attorneys, and
patients "expect" them to utilize such technologies lavishly in order for them to be considered "good"
doctors. That idea is nonsense.
An often-ignored aspect of medical practice is that of biostatistics, and within that
topical area there resides a concept known as the "rule of prior
probability." It states that if one is already absolutely sure of a conclusion concerning any
given question, the procurement of additional data with a bearing on that question cannot possibly
increase the level of certitude—it can only produce confusion. A common example is represented by the
classical image of small-cell carcinoma in a lung biopsy. Rather than simply making the diagnosis, the
pathologist obtains several immunostains for "neuroendocrine" markers, which return negative results. He
or she is then confused as to a diagnostic plan of action; the morphologic picture is certain, but the
DIHC data are obfuscatory. In short, the latter studies should not have been obtained in the first place
in that specific context.
Another murky topic concerns the definition of a "positive" immunostain. From mysterious
sources unknown to me, published requirements for percentages of positive cells and "strengths" (1+, 2+,
etc.) of immunoreactivity have emerged as stipulations for "positivity," but all of them are
eminently-subjective if not confabulated criteria that are completely dependent on the details of
technical procedure. The only worthwhile tenet attending this subject is one emphasized by Dr. Rod
Miller; he has stated that "true positive" immunostains exhibit cell-to-cell
heterogeneity in labeling, whereas "pseudo-positive" reactions are diffusely homogeneous.
What guidelines can be offered, then, with regard to the ordering and interpretation of
DIHC studies? They are simply the following:
1. Base the selection of stains on a morphologically-sound & suitably-narrow differential diagnosis;
2. Choose only
those studies that will provide a high level of diagnostic discrimination between the
pathologic entities being considered—the fewer stains the better;
3. Never put
great weight on a single immunostaining result if it blatantly contradicts the bulk of the other
clinicopathologic data in the case at hand;
4. Do not use the results of technically-equivocal immunostains to
contribute to a diagnostic conclusion.
Returning to the exemplary seminar case of cellular neurothekeoma (CN), the only potential "clue" to a
possible problem with the original set of immunostains was the pattern of reactivity for S100 protein.
It demonstrated perfectly homogeneous and nuclear-only labeling, which, in
retrospect, was probably an indicator of "pseudo"-positivity. What the particular
cause of that aberrant staining was, with respect to
the several potential factors discussed earlier in this précis, I cannot say. I will also freely
acknowledge that on a busy day, given a healthy case-load, I would not be at all certain of pouncing on
that S100 protein stain as a problem. C'est la vie.
CN is a particularly troublesome tumor immunohistologically, because, as a neoplastic
entity, it is still so nebulously defined. Proposals have been made suggesting that CN is a
neural-neuroendocrine, or melanocytic, or "fibrohistiocytic," or hybrid lesion with respect to its
lineage of differentiation. Phenotypically, it may be reactive for muscle-specific actin; NKI/C3 (a
melanocyte-related marker); microophthalmia-transcription factor (another melanocyte-related moiety);
protein gene product 9.5 (a neural/neuroendocrine marker); neuron-specific enolase; p63; CD10;
podoplanin; and S100A6, a specific calcium-binding S100 isoform. However, it consistently lacks generic
S100 protein-positivity, as well as HMB45, melan-A, KBA-62 antigen, PNL2 protein, and tyrosinase, as seen
in true melanomas. Just as problematically, only 90% of melanomas express the latter markers in addition
to S100 protein; they also consistently manifest S100A6, and may demonstrate unexpected
immunoreactivities (e.g., for myogenous, neural, or epithelial determinants)
because of the presence of divergent differentiation.
Hence, the crux of the seminar case centers on the following question—was it an "error" to
make the diagnosis of melanoma in this instance, rather than one of CN? The answer depends on one's
point of view. Dr. Ronald Sirota, an authority on medical liability, has stated that "…Assignment of the term 'error' is a human judgment…and this judgment can be flawed for
a number of reasons." Foucar has aptly framed the situation that pertains to CN, in saying that
"…If one concludes that anatomic pathology currently lacks precise or accurate
criteria relating to certain diagnoses, a situation exists that is comparable to a lack of effective
therapy for a particular disease… Yearning for a definitive pathologic diagnosis does not substitute for
reality when no method to achieve that interpretation can be found." Thus, if no incontrovertible
procedure exists for recognizing CN, the diagnostic vagaries of immunohistology become somewhat
irrelevant to the case in point. However, that should not detract us from keeping them in our fund of
knowledge as potential pitfalls in a general context.
Alkhalidi H, Ghazarian D. Cellular neurothekeoma with a plexiform morphology:
a case report with a discussion of the plexiform lesions of the skin. J Cutan
Pathol 2007; 34: 264-269.
Banerjee SS, Eyden B. Divergent differentiation in malignant melanomas: a
review. Histopathology 2008; 52: 119-129.
Barnhill RL, Mihm MC Jr. Cellular neurothekeoma. A distinctive variant of neurothekeoma mimicking
nevomelanocytic tumors. Am J Surg Pathol 1990; 14: 113-120.
Fetsch JF, Laskin WB, Hallman JR, Lupton GP, Miettinen M. Neurothekeoma: an
analysis of 178 tumors with detailed immunohistochemical data and long-term patient follow-up
information. Am J Surg Pathol 2007; 31: 1103-1114.
Foucar E: Classification of error in anatomic pathology: a proposal for an
evidence-based standard. Semin Diagn Pathol 2005; 22: 139-146.
Hornick JL, Fletcher CD. Cellular neurothekeoma: detailed characterization in
a series of 133 cases. Am J Surg Pathol 2007; 31: 329-340.
Jaffer S, Ambrosini-Spaltro A, Mancini AM, Eusebi V, Rosai J. Neurothekeoma and plexiform
fibrohistiocytic tumor: mere histologic resemblance or histogenetic relationship? Am J Surg Pathol 2009; 33: 905-913.
Jo VY, Fletcher CD. p63 immunohistochemical staining is limited in soft tissue
tumors. Am J Clin Pathol 2011; 136: 762-766.
Kaddu S, Leinweber B. Podoplanin expression in fibrous histiocytomas and cellular neurothekeomas.
Am J Dermatopathol 2009; 31: 137-139.
Leclerc-Mercier S, Brousse N, Fraitag S. Is plexiform fibrohistiocytic tumor a deep form of cellular
neurothekeoma? J Cutan Pathol 2009; 36: 1123-1125.
Miller RT: True positive vs. false positive staining. http://www.propath.com/index2.php?option=com_content&do-pdf=1&id=148
Plaza JA, Torres-Cabala C, Evans H, Diwan AH, Prieto VG. Immunohistochemical
expression of S100A6 in cellular neurothekeoma: clinicopathologic and immunohistochemical analysis of
31 cases. Am J Dermatopathol 2009; 31: 419-422.
Sirota RL: Error and error reduction in pathology. Arch Pathol Lab
Med 2005; 129: 1228-1233.
Vered M, Fridman E, Carpenter WM, Buchner A: Classic neurothekeoma (nerve sheath myxoma)
and cellular neurothekeoma of the oral mucosa: immunohistochemical profiles. J
Oral Pathol Med 2011; 40: 174-180.
Wartchow EP, Goin L, Schreiber J, Mierau GW, Terella A, Allen GC. Plexiform
fibrohistiocytic tumor: ultrastructural studies may aid in discrimination from cellular
neurothekeoma. Ultrastruct Pathol 2009; 33: 286-292.
Wick MR, Mills NC, Brix WK: Tissue procurement, processing, & staining techniques.
In: Diagnostic Histochemistry (Wick MR, Ed),
Cambridge University Press, New York, 2008; pp. 1-27.
Wick MR, Swanson PE, Marchevsky AM: Evidence-based practices in applied immunohistochemistry:
dilemmas caused by cross-purposes. In: Evidence-Based
Pathology & Laboratory Medicine (Marchevsky AM, Wick MR, Eds), Springer, New York,
2011; pp. 261-295.