CytoLyt® Fixation Significantly Inhibits MIB1 Immunoreactivity, whereas Alternative Ki67 Clone 30-9 is not Susceptible to the Inhibition: Critical Diagnostic Implications

Abstract

Background

Ki67 proliferation marker has multiple diagnostic and prognostic applications. While several clones to Ki67 antigen are commercially available, MIB1 clone is widely recommended in surgical pathology literature for neuroendocrine tumors. In our cytopathology practice, we have encountered unexpectedly low MIB1 immunoreactivity in CytoLyt®-fixed cell blocks (CB). This study evaluated the impact of fixatives, CB processing and immunocytochemical (ICC) procedures on Ki67 immunoreactivity.

Methods

Test CBs were prepared from freshly-resected tumors, and multiple variables in MIB1 ICC procedure were tested, including CytoLyt® versus formalin collection media, MIB1 versus other Ki67 clones including 30-9 (Ventana), and other variables. MIB1 versus 30-9 clones were tested in parallel on CytoLyt®-fixed CBs from clinical samples of small cell lung carcinomas (SCLC).

Results

In test CBs (n=10), mean MIB1 labeling index was 10% in CytoLyt® versus 47% in formalin (P=0.0116), with mean loss of reactivity in matched CBs of 37% (up to 70%). None of the procedure modifications tested in 223 individual ICC reactions recovered MIB1 reactivity in CytoLyt® except for switching to 30-9 antibody. In CytoLyt®-fixed SCLC samples (n=14), Ki67 30-9 exhibited expected ranges of reactivity (mean 83%, range 60–100%), whereas MIB1 showed markedly inhibited labeling (mean 60%, range 10–95%; P=0.0058).

Conclusions

CytoLyt® fixation substantially inhibits MIB1 immunoreactivity, whereas Ki67 30-9 clone is not susceptible to inhibition. Markedly discrepant MIB1 reactivity may present a pitfall in the diagnosis of SCLC and may lead to incorrect prognostic stratification of other tumor types. For laboratories utilizing CytoLyt®, we recommend using Ki67 30-9 rather than MIB1 antibody.

Precis:

CytoLyt® fixation substantially inhibits MIB1 immunoreactivity and may present a diagnostic pitfall in certain tumors, especially small cell carcinoma. The Ki67 30-9 clone shows adequate performance and we strongly recommend it for laboratories utilizing CytoLyt® collection media.

BACKGROUND

Quantitative assessment of Ki67 proliferation marker immunoreactivity is a well-established and widely-utilized ancillary tool in tumor diagnosis and prognostication.¹ Assessment of Ki67 proliferation index is used in pathology practice to stratify neuroendocrine tumors as well as other solid tumors and certain lymphomas.^2–6 While several antibody clones to Ki67 antigen are currently commercially available, the one that is currently widely recommended in surgical pathology literature for the assessment of neuroendocrine tumors is MIB1.⁷

In patients with metastatic disease, cytology specimens are commonly preferred to surgical biopsies as they are less invasive and more cost-effective diagnostic methods. Cell blocks coupled with immunocytochemistry (ICC) are essential for this process regardless of whether a laboratory utilizes formalin, CytoLyt® or other fixatives as collection media. The advantage of collecting cell blocks in alcohol-based media like CytoLyt® (methanol-based medium) or SurePath (ethanol-based medium) is that they allow preparation of liquid-based monolayers, which allow excellent visualization of cytomorphology. In addition, alcohol-based fixatives are superior to formalin for molecular studies because alcohol does not cross-link DNA.⁸ Generally, however, immunostains are validated on formalin-fixed paraffin embedded (FFPE) tissue, whereas the performance of antibodies on CytoLyt®-fixed cell blocks is often uncertain and some stains, such as S100 and ER, have been reported to perform suboptimally in such preparations.⁹

In our cytopathology practice, until recently, we have utilized CytoLyt® collection media. After adopting MIB1 antibody following the current recommendations, we started encountering unexpectedly low MIB1 immunoreactivity in CytoLyt®-fixed cell blocks, as compared to the immunoreactivity seen in the subsequent formalin-fixed resections or concurrent core biopsy specimens. This study was designed to examine the factors contributing to unexpectedly low reactivity, including comparison of fixatives (CytoLyt® versus formalin), antibodies (MIB1 vs other commercially available clones including 30-9 from Ventana), as well as a number of other cell block processing and ICC procedures. To directly assess potential impact of inhibited MIB1 reactivity in clinical practice, we analyzed the performance of MIB1 vs Ki67 30-9 in cells blocks from clinical samples of small cell lung carcinoma (SCLC) – a tumor type known to have consistently high Ki67 proliferation index, where erroneous proliferation rate assessment can have critical diagnostic implications.

METHODS

Freshly resected carcinomas of various types (n=10) were gently scraped at the time of specimen prosection to prepare test cell blocks. Each scrape was divided evenly into CytoLyt® and 10% neutral buffered formalin, and a subset of tumors was also split in parallel into formol-alcohol (9 parts 95% EtOH : 1 part 40% formalin). Each container was then processed using the standard cell block preparation method in our cytology laboratory (HistoGel+ethanol).¹⁰ A set of cell blocks was also prepared using the HistoGel-only method, excluding the ethanol step. A total of 38 test cells blocks were prepared (Supplementary Table 1).

Immunocytochemistry for Ki67 antigen was performed employing monoclonal antibody (mAB) MIB1 (DAKO/Agilent, Santa Clara, CA; M7240) using a standard protocol utilized for histological specimens: 1:200 dilution, antigen retrieval time 40 min, CC1 retrieval, incubation time 32 min on Ventana Benchmark Ultra platform (Ventana, Tuscon, AR). Performance of this protocol was compared for cell blocks collected in parallel in CytoLyt® vs Formalin vs Formol-alcohol. The following modification in ICC protocol were tested with MIB1 antibody: different antigen retrieval methods (CC1 vs ER1 vs ER2), different retrieval times (10–25 mins), increased DAKO antibody concentration (1:50), and on Leica-Bond-3 platform. Lastly, we compared the performance of MIB1 to other Ki67 clones - 30-9 (790–4286; Ventana, RTU, CC1, 40 min) and D2H10 (#9270; Cell Signaling Technologies, Danvers, MA) at various dilutions (1:800 & 1:1500). See Supplementary Table 2 for full summary of tested condition.

To compare the performance of MIB1 and Ki67 30-9 in clinical samples of SCLC, we identified 14 fine needle aspirates (FNAs) with the diagnosis of SCLC (10 EBUS-FNAs and 4 percutaneous FNAs), for which cellular CytoLyt®-fixed cell blocks were available. Recuts from those cells blocks were stained in parallel with MIB1 (DAKO, 1:200, ER2, 30 min) and Ki67 30-9 (Ventana, RTU, CC1, 40 min) antibodies.

Cell blocks with scant amount of viable tumor cells (less than approximately 1 thousand cells) were excluded both for test cell blocks from scrapes and clinical small cell carcinoma cell blocks. For CBs with sufficient cellularity, percentage of positive tumor staining was carefully estimated as is routinely done in our clinical practice by the same cytopathologist to ensure consistency. The percentage of positive tumor cells was compared using two-tailed unpaired t test.

RESULTS

Ex-vivo cell blocks: MIB1 in CytoLyt® vs Formalin vs Formol-Alcohol

MIB1 (DAKO, 1:200, ER2, 32 min) immunoreactivity was compared in test cell blocks prepared from split ex-vivo tumor scrapes fixed in parallel CytoLyt® versus formalin from 10 tumor resections. Immunoreactivity in the CytoLyt®-fixed cell blocks was consistently lower than that in matched formalin-fixed cell blocks: mean tumor cell labeling in formalin was 47% compared to 10% in matched cell blocks fixed in CytoLyt® [P=0.0116], with a mean loss of reactivity in matched cell blocks of 37% (up to 70%) (Figure 1). In addition, MIB1 staining intensity was weaker in CytoLyt®-fixed cell blocks (1–2+ in nearly all cases) compared to those fixed in formalin (3+ in all cases). Almost complete loss of MIB1 staining was observed in formol-alcohol.

Figure 1:

A. MIB1 immunoreactivity in test cell blocks from 10 resected tumors scraped in parallel into Formalin vs CytoLyt®. B. Dot plots in which each dot represents immunoreactivity in a single sample. Bars show mean and standard deviation.

Ex-vivo cell blocks: Testing other antibodies, various modifications of cell block processing and ICC protocols

We tested multiple variations in the MIB1 ICC protocol, antigen retrieval, and cell block processing with the attempt to recover MIB1 reactivity in CytoLyt®. A total of 223 individual immunostains were performed, as summarized in Supplementary Table 2. None of the tested modifications (see Methods and Supplementary Table 2) restored immunoreactivity of MIB1 in CytoLyt® to the level seen with formalin fixation.

Remarkably, we found that a different Ki67 clone – 30-9 (Ventana) – showed significantly superior reactivity compared to MIB1 in CytoLyt®. In all cases, tumor cells showed greater percentage of positive tumors cells and consistently higher intensity (2–3+ in all cases) (Figure 2). Conversely, another commercially-available Ki67 antibody - D2H10 (Cell Signaling) – was also markedly inhibited in CytoLyt® with multiple tested protocol conditions.

Figure 2:

Ki67 30-9 antibody has superior performance compared to MIB1 antibody in CytoLyt®-fixed cell blocks. Shown are examples of reactivity for MIB1 (B, E) versus Ki67 30-9 (C, F) in CytoLyt®-fixed cell blocks from an endometrioid adenocarcinoma (case 1) and carcinosarcoma (case 2).

Clinical samples of small cell carcinoma: MIB1 vs. Ki67 30-9

Parallel staining with MIB1 versus Ki67 30-9 was performed on CytoLyt®-fixed cell blocks from 14 clinical cases of SCLC. Ki67 30-9 clone exhibited expected ranges of reactivity (mean 83%, range 60–100%), whereas MIB1 showed markedly lower labeling (mean 60%, range 10–95%) [P=0.0058; Figure 3 and Figure 4]. On average, MIB1 rate was 23.6% (up to 70%) lower compared to Ki67 in matched samples. In 5 cases, MIB1 labeling was markedly discrepant, showing labeling in <50% of cells. In addition, while the intensity of staining with 30-9 was 3+ in almost all cases, MIB1 showed substantially lower intensity (1–2+ in the majority of cases).

Figure 3:

A. Immunoreactivity for Ki67 30-9 vs MIB1 antibodies in CytoLyt®-fixed cell blocks of small cell lung carcinoma. B. Dot plots in which each dot represents immunoreactivity in a single sample. Bars show mean and standard deviation.

Figure 4:

Example of markedly inhibited proliferation index with MIB1 compared to 30-9 clones of Ki67 antibodies in CytoLyt®-fixed cell blocks from small cell lung carcinoma.

DISCUSSION

Proliferation rate as assessed by immunostaining for Ki67 antigen is widely utilized for diagnostic and prognostic applications in various tumors.^11–14 One area where it is of importance is in the diagnosis and grading of neuroendocrine neoplasms. Diagnostically, Ki67 has a critical important role in distinguishing well-differentiated neuroendocrine tumors (such as lung carcinoids) from high-grade carcinomas (small cell and large cell neuroendocrine carcinoma) in small biopsies and cytology specimens. In cytology specimens, the distinction of these tumors types may be challenging. Prior studies have shown that Ki67 proliferation index is an effective ancillary tool for distinguish these tumors, with proliferation index of up to 20% consistent with carcinoid tumors, and of 50–100% supporting high grade neuroendocrine carcinomas.^15,16 Furthermore, the precise level of Ki67 is used to grade gastro-entero-pancreatic neuroendocrine tumors (G1 <3%, G2 3 – 20% and G3 >20%), which is used to predict prognosis and guide treatment selection.¹⁷ Other tumor types where quantitative assessment of Ki67 is used in routine practice includes non-Hodgkin lymphoma, especially mantle cell lymphoma, and a variety of other solid tumors including breast carcinomas and meningiomas.^11,12,14,18 Though MIB1 and 30-9 clones have been shown to have a high concordance in formalin-fixed material,¹⁹ MIB1 is the one that is recommended in neuroendocrine tumor literature^6,7 given that the key studies on the prognostic significance of Ki67 proliferation index were performed using this clone.¹⁷ Our findings therefore have important implications since MIB1 staining in CytoLyt®-fixed cell blocks can yield markedly under-estimated values. This can lead to incorrect diagnosis, for example if the differential diagnosis is between carcinoid tumor and SCLC, or to marked under-grading of a gastro-entero-pancreatic neuroendocrine tumors.

Antibodies are typically validated for use in the formalin-fixed paraffin embedded tissue, whereas validation may be lacking for material fixed in other fixatives such as CytoLyt®. CytoLyt®, a methanol-water solution, is a common collection media for cytology specimens (fluids, exfoliative specimens and FNAs). It is well documented that a number of antibodies perform sup-optimally in methanol-based fixatives compared to FFPE tissue, although specific antibodies affected vary in different studies, likely depending on specific protocols and laboratory conditions utilized in different institution.^9,20–22 One study did mention MIB1 as one of affected antibodies,²⁰ but overall, this issue is not well known in cytopathology practice. The reason that suboptimal reactivity with this antibody is particularly important is that unlike most diagnostic antibodies in pathology, which are not quantitative and for which some inhibition of labeling may not lead to major diagnostic implications, the interpretation of Ki67 depends on precise quantitative cut-offs. These are significantly discrepant in CytoLyt®-fixed cell blocks.

Although we cannot exclude the possibility that certain staining conditions used in our laboratory had exacerbated the degree of inhibition in CytoLyt®-fixed cell blocks, we have also encountered inhibition of MIB1 in CytoLyt®-fixed blocks in specimens submitted to our institution from various other laboratories, and which were corrected by using 30-9 antibody. Thus, this problem is not limited to the procedures of our laboratory.

One seemingly simple solution to the described problem with CytoLyt® would be to post-fix the specimens in formalin. However, in out pilot studies we found that post-fixation in formalin does not result in restoration of immunoreactivity (data not shown). Likewise, Hanley et al showed that post-fixation in formalin did not recover inhibition of progesterone receptor immunoreactivity in alcohol-fixed cell blocks.²³ It is likely that the effects of alcohol on antigen configuration cannot be reversed after the exposure to the initial collection media.

After recognizing that Ki67 30-9 is superior to MIB1 in CytoLyt® fixed cell block, we have incorporated using Ki67 30-9 into our clinical practice. We have found it to be particularly useful in diagnosing SCLC of lung origin where it is common that the only material available is a cytology specimen obtained from an FNA procedure. To illustrate the potential impact of MIB1 inhibition on clinical samples of SCLC, here we performed parallel stains for MIB1 and 30-9 on a series of SCLC collected in CytoLyt®. We observed lower proliferation indices with MIB1 antibody in the majority of cases. Importantly, in some cases MIB1 rate fell below 50% (in one case down to 10%). Rates of <50% are extremely uncommon in SCLC and could therefore lead to misinterpretation of SCLC as a carcinoid tumor. In fact, a challenge of this type was encounter in several neuroendocrine tumors in our laboratory before we were aware of this problem, and we continue to encounter this issue in specimens submitted to us from other laboratories.

We noted that there is some variability in the MIB1 inhibition in clinical samples collected in CytoLyt®, in that in some cases staining is inhibited drastically, while in other cases inhibition is minimal or absent compared to 30-9. We hypothesize that this variability may be relate to the length of fixation in CytoLyt® or individual case variables in specimen collection or fixation.

Following the results of this study, as well as increasing recognition of the effects of CytoLyt® on the performance of multiple antibodies, our laboratory has switched to collecting FNA needle rinses in formalin rather than CytoLyt®. However, some effusion fluids and exfoliative specimens may still be collected in CytoLyt® media. Furthermore, according to the recent survey, a significant number of US laboratories utilize CytoLyt® collection media.²⁴ Our findings are therefore relevant for a large number of laboratories.

Notably, to our knowledge, there have been no technical issues documented in the literature for MIB1 antibody in formalin-fixed histological specimens. Thus, preferentially using the 30-9 antibody applies only to CytoLyt®-fixed cell blocks, although detailed comparative studies of antibody performance in histologic specimens would be of value.

CONCLUSION

Here we document that the MIB1 antibody is markedly inhibited by CytoLyt® fixation, whereas the alternative clone – Ki67 30-9 – is not susceptible to this inhibition. We therefore recommend using Ki67 30-9 antibody for CytoLyt®-fixed cells blocks, especially in tumors where quantitative assessment of proliferation index has significant diagnostic or clinical implications.