Abstract
We compared grading systems and examined associations with tumor stroma and survival in patients with cervical squamous cell carcinoma (SCC).
Available tumor slides were collected from 10 international institutions. Broders tumor grade, Jesinghaus grade (informed by the pattern of tumor invasion), Silva pattern, and tumor stroma were retrospectively analyzed; associations with overal survival, progression-free survival, and presence of lymph node metastases were examined. Binary grading systems incorporating tumor stromal changes into Broders and Jesinghaus grading systems were developed.
Of 670 cases, 586 were reviewed for original Broders tumor grade, 587 for consensus Broders grade, 587 for Jesinghaus grade, 584 for Silva pattern, and 556 for tumor stroma. Reproducibility between grading systems was poor (κ=0.365, original Broders/consensus Broders; κ=0.215, consensus Broders/Jesinghaus).
Median follow-up was 5.7 years (range, 0–27.8). Progression-free survival rates were 93%, 79%, and 71%, and overall survival rates were 98%, 86%, and 79% at 1, 5, and 10 years, respectively. On univariable analysis, original Broders (P<0.001), consensus Broders (P<0.034), and Jesinghaus (P<0.013) grade were significant for overall survival; original Broders grade was significant for progression-free survival (P=0.038). Predictive accuracy for overall survival and progression-free survival were 0.559 and 0.542 (original Broders), 0.542 and 0.525 (consensus Broders), 0.554 and 0.541 (Jesinghaus grade), and 0.512 and 0.515 (Silva pattern), respectively. Broders and Jesinghaus binary tumor grades were significant on univariable analysis for overall survival and progression-free survival, and predictive value was improved. Jesinghaus tumor grade (P<0.001) and both binary systems (Broders, P=0.007; Jesinghaus, P<0.001) were associated with presence of lymph node metastases.
Histologic grade has poor reproducibility and limited predictive accuracy for SCC. The proposed binary grading system offers improved predictive accuracy for survival and presence of lymph none metastases.
Keywords: cervix, squamous cell carcinoma, prognosis, histologic grading, stroma
Introduction
Cervical cancer is one of the most common cancers among women worldwide, especially in low- and middle-income countries. Seventy-five percent of all cervical cancers are squamous cell carcinomas (SCCs); the remaining 25% are a combination of endocervical adenocarcinoma, adenosquamous carcinoma, and other rare subtypes.1
Pathologists currently use various methods to grade SCCs, such as Broders classification as well as tumor invasion pattern, which was proposed by Jesinghaus et al.2–4 The Broders classification system was proposed in 1920, and since then, it has been adopted in many centers.2 In contrast to other clinicopathologic parameters, such as International Federation of Gynecology and Obstetrics (FIGO) stage, tumor diameter, depth of invasion, presence of lymphovascular invasion, and lymph node metastases, which are well-known prognostic factors in SCC, the prognostic and predictive significance of histologic tumor grade is debatable.5–7 Due to its uncertain clinical and prognostic value, tumor grade is not recommended for pathology report inclusion by the 2020 World Health Organization classification system or by other guidelines, including those from the International Collaboration on Cancer Reporting (ICCR; non-core element, only supported by III-2 level of evidence), the European Society of Gynaecological Oncology, the European Society for Radiotherapy and Oncology, the Gynecologic Oncology Group, and the European Society of Pathology.8–10
Previous studies have found conflicting results regarding the relationship between tumor stromal reaction and survival in SCCs. Some studies have demonstrated that the presence of a stromal reaction to tumor invasion, and specifically a myxoid tumor stroma compared to fibrous stroma, is associated with worse prognosis and decreased survival in cervical, vulvar, and endometrial cancers.11,12 In one study, a desmoplastic stromal response was more often observed with an aggressive spray-like pattern of invasion, which was significantly associated with poorly differentiated tumors (particularly in advanced stages) and reduced overall survival;13,14 however, this has not been extensively studied in association with various histologic grading systems of SCC.
Broders classification was evaluated in a multivariate analysis along with other prognostic variables in an article (accepted for publication) by our group, which found Broders grade did not have independent prognostic associations with clinical outcomes. This finding prompted us to explore whether other grading systems might be more powerful than Broders grade.
In the present study, we compared various grading systems (Broders classification, the grading system proposed by Jesinghaus et al, and Silva pattern of invasion), and assessed their reproducibility and relation to overall survival and progression-free survival, in a large, multi-institutional, international cohort of patients with SCC. We also investigated the association of these grading systems with presence of lymph node metastases in order to identify which patients may benefit from extensive surgery. To improve morphology-based grading and guide development of a new binary grading system, we investigated various grading systems in combination with tumor stromal assessment.
Materials and Methods
Original tumor slides from patients with surgically treated SCC were collected from 10 international institutions and retrospectively analyzed (University of Medicine, Pharmacy, Sciences and Technology “George E Palade” of Targu Mures, Romania [8/19/2011–6/5/2015]; Memorial Sloan Kettering Cancer Center, New York, NY, USA [5/18/1992–7/2/2021]; Vancouver General Hospital, BC, Canada [2/1/2017–12/6/2019]; Regional Institute of Oncology, Iasi, Romania [1/28/2012–9/10/2018]; Ospedale Sacro Cuore Don Calabria, Negrar, Italy [11/27/1996–7/10/2019]; Sahlgrenska University Hospital, Gothenburg, Sweden [11/24/2014–12/12/2017]; University of Chicago, Chicago, IL, USA [4/22/2010–5/10/2017]; Jikei University School of Medicine, Tokyo, Japan [6/10/2009–12/7/2018]; Wayne State University, Detroit, MI, USA [5/29/2002–1/10/2019]; and Massachusetts General Hospital, Boston, MA, USA [1/24/2001–3/5/2018]). Inclusion criteria comprised the following: patients with SCC confirmed on microscopic examination; surgical treatment with cone/loop electrosurgical excision procedure/trachelectomy/hysterectomy/ exenteration with or without lymph node dissection; no oncologic treatment prior to surgery; all hematoxylin and eosin original slides from the cervical tumor were available; access to all clinical, pathologic, and follow-up data; and Institutional Review Board approval from the treating institution. Exclusion criteria included the following: patients with SCC confirmed only on biopsy; microscopic examination revealed another subtype of cervical carcinoma (eg, adenocarcinoma, adenosquamous); and treatment with chemotherapy and/or radiotherapy prior to surgery.
Tumors were initially classified and graded by pathologists at the originating institutions (original Broders tumor grade). This grade was usually based on institutional or individual approximates of the Broders grade. Two pathologists (SS and RAS) reviewed all cases (with a median of 12 slides per case) and graded them using the modified Broders classification system, accounting for degree of keratinization, cytologic atypia, mitotic activity, and more specific cutoffs;2,15 this was designated the consensus Broders tumor grade (Figure 1). In well-differentiated tumors (grade 1), numerous keratin pearls were observed in the center of tumor cell nests; tumor cells appeared mature, with abundant eosinophilic cytoplasm, intercellular bridges and large but uniform nuclei, and few mitotic figures (<3 mitotic figures per 10 high-power fields [HPFs]). In moderately differentiated tumors (grade 2), keratin pearls were rare, neoplastic cells showed more pleomorphic nuclei (even focally), with less abundant cytoplasm, indistinct intercellular bridges, and more mitotic figures (3–10 mitotic figures per 10 HPFs). Poorly differentiated tumors (grade 3) were composed of large tumor cells with hyperchromatic and very pleomorphic nuclei, scant cytoplasm (resembling tumor cells of high-grade squamous intraepithelial lesion), and numerous abnormal mitotic figures (>10 mitotic figures per 10 HPFs); clear-cut squamous differentiation manifested by keratinization was absent (Table 1). The tumor growth pattern informed the Jesinghaus tumor grade and the Silva pattern of invasion, which were evaluated along with tumor stroma by the same pathologists (SS and RAS).
Figure 1:
Cervical squamous cell carcinoma (SCC) based on Broders classification.
(A) Well-differentiated, grade 1; (B) moderately differentiated, grade 2; and (C) poorly differentiated, grade 3.
Table 1:
Consensus Broders grading system
| Grade | Presence of keratin pearls | Tumor cells | Mitotic figures |
|---|---|---|---|
| Grade 1 (well-differentiated) | Numerous | Mature, with abundant eosinophilic cytoplasm, intercellular bridges and large but uniform nuclei | <3 mitotic figures per 10 HPF |
| Grade 2 (moderately differentiated) | Rare | Pleomorphic nuclei (even focally), with less abundant cytoplasm, indistinct intercellular bridges | 3–10 mitotic figures per 10 HPF |
| Grade 3 (poorly differentiated) | Very rare/Absent | Hyperchromatic and very pleomorphic nuclei, scant cytoplasm | >10 mitotic figures per 10 HPF |
HPF, high-power fields
The grading system proposed by Jesinghaus et al in 2018 incorporates the number and size of tumor buds.3 Tumor budding activity was evaluated throughout the entire tumor area and scored based on highest budding activity (hot spot) in 10 HPFs. Tumor budding activity was classified as low and high if 1–14 or ≥15 budding foci were detected in 10 HPFs, respectively. Cell nests composed of >15 and 5–15 tumor cells were classified as large and intermediate, respectively. Cell nests consisting of 2–4 tumor cells were scored as small, while single tumor-cell invasion was reserved for tumors harboring singular, discohesive tumor cells without nested architecture, detached within the stroma. The smallest identifiable cell nest size was reported for every case. Consequently, the Jesinghaus grading system was based on the sum of budding activity and the smallest identifiable cell nest size, attributing a score to both budding activity (1–3 points) and cell nest size (1–4 points). Grade 1 tumors have a score ranging from 2 to 3, grade 2 tumors have a score ranging from 4 to 5, and grade 3 tumors have a score ranging from 6 to 7 (Table 2) (Figure 2).3
Table 2:
Jesinghaus grading system
| Grade | Number of buds* | Size of buds | Novel grade score** |
|---|---|---|---|
| 1 | 0 | >15 tumor cells (large) | 2–3 score: grade 1 (G1, well differentiated) |
| 2 | 1–14 | 5–15 tumor cells (intermediate) | 4–5 score: grade 2 (G2, moderately differentiated) |
| 3 | >15 | 2–4 tumor cells (small) | 6–7 score: grade 3 (G3, poorly differentiated) |
| 4 | - | single tumor cell detached within the stroma |
The number of buds are determined in the highest budding activity (hotspot), in 10 high-power fields (HPFs)
The sum of tumor budding and cell cluster size score generates a final grading score ranging from 2 to 7
Figure 2:
Jesinghaus tumor grade. (A) Low tumor budding activity and large nests; (B) Overview on high budding activity; (C) Intermediate and small tumor cell nests detached within the stroma; (D) Various sizes of tumor buds, including single-tumor cell invasion; and (E) ≥15 budding foci (high tumor budding activity).
The Silva pattern of invasion system, initially designed for endocervical adenocarcinomas, was also applied in each case, separating tumors into patterns A, B, or C based on absence or presence and size of stromal destructive invasion and presence or absence of lymphovascular invasion (Table 3) (Figure 3).16
Table 3:
Silva system based on pattern of invasion and presence of lymphovascular invasion (LVI)
| Pattern | Morphologic parameter | LVI status |
|---|---|---|
| Silva A pattern | Usually preserved architecture; well-demarcated glands, rounded contour; no destructive stromal invasion | No LVI |
| Silva B pattern | Similar architecture to pattern A at low power magnification; early destructive stromal invasion present (less than 4 x field, 5 mm in diameter | LVI may be present |
| Silva C pattern | High nuclear grade; extensive and diffuse destructive stromal invasion | LVI may be present |
LVI, lymphovascular invasion
Figure 3:
Silva pattern of invasion in cervical squamous cell carcinoma (SCC). (A) Silva A pattern; (B) Silva B pattern; and (C) Silva C pattern.
We developed two binary grading systems by incorporating the nature of tumor stromal changes into Broders and Jesinghaus tumor grading systems, respectively (Figure 4). Tumor stroma was considered fibrous when pink in color and represented by compact proliferation of fibroblasts. In contrast, tumor stroma was considered myxoid when myxo-edematous (of light-blue color, basophilic) and associated with inflammatory infiltrate. Myxoid stroma was scored in areas with the highest bud counts that were easily identified (in more than one focus) at scanning magnification (Figure 5). To create the novel binary grading systems, Broders grade 2 carcinomas with fibrous stroma were grouped with cases of Broders grade 1 (Broders binary low grade), while cases of Broders grade 2 with myxoid stroma were grouped with cases of Broders grade 3 (Broders binary high grade). Similarly, cases of Jesinghaus grade 2 with fibrous stroma were grouped with cases of Jesinghaus grade 1 (Jesinghaus binary low grade), while cases of Jesinghaus grade 2 with myxoid stroma were grouped with cases of Jesinghaus grade 3 (Jesinghaus binary high grade) (Figure 4, Table 4).
Figure 4:
Broders and Jesinghaus binary grading systems: low and high grade.
Figure 5:
Cervical squamous cell carcinoma (SCC) with (A) fibrous stroma and (B) myxoid stroma.
Table 4:
Various grading systems applied in the present study
| Type of grade | Parameters | Tumor grade |
|---|---|---|
| Broders tumor grade | Degree of keratinization, cytologic atypia and mitotic activity | Grade 1, 2, 3 |
| Jesinghaus tumor grade | Number and size of tumor buds | Grade 1, 2, 3 |
| Broders binary tumor grade | Broders system and stroma | Broders binary low grade Broders binary high grade |
| Jesinghaus binary tumor grade | Jesinghaus system and stroma | Jesinghaus binary low grade Jesinghaus binary high grade |
Information related to the presence of lymph node metastases, follow-up, and survival was retrieved from clinical and pathological records at each institution. Categorical variables were summarized using frequency and percentages. Progression-free survival was defined as the time from surgery to progression or death, and overall survival was defined as the time from surgery to last follow-up or death. Overall survival and progression-free survival were evaluated using Kaplan-Meier methodology, and univariable Cox proportional hazards models clustered by institution were created to determine associations with grading types. Univariable logistic regression models were used to assess the associations of grading types with presence of lymph node metastases. Kappa statistics were generated to determine agreement between the different grading types, and concordance probability estimates (CPEs) were used to assess the predictive accuracy of each grade for overall survival and progression-free survival.17 Sensitivity analyses were conducted to determine if clinical variables were associated with an upgrade or downgrade across the grading systems, and chi-square, Fisher exact, and Kruskal-Wallis tests were used to evaluate the changes. A P value <0.05 was considered statistically significant. All analyses were conducted in R v.4.2.2 (https://www.R-project.org/).18
Results
A total of 670 patients with SCC were identified and included in the study. The original Broders tumor grade was evaluated in 586 cases, consensus Broders and Jesinghaus tumor grade were evaluated in 587 cases, and Silva pattern in 584 cases. Original Broders tumor grade was not evaluated by the submitting institution in 84 cases, per institutional protocol. Due to missing paraffin blocks/slides, consensus Broders tumor grade and Jesinghaus tumor grade were not evaluated in 83 cases, and 86 cases did not have a secondary review for Silva pattern of invasion. Tumor stroma was not evaluated in 114 cases.
Twenty-six cases (4%) were original Broders tumor grade 1, 340 (58%) were grade 2, and 220 (38%) were grade 3. Seventeen cases (3%) were consensus Broders grade 1, 356 (61%) were grade 2, and 214 (36%) were grade 3. Jesinghaus grade 1 was found in 25 cases (4%), grade 2 in 260 cases (44%), and grade 3 in 302 cases (51%). Silva pattern of invasion was A in 15 cases (3%) and B/C in 569 cases (97%) (Table 5).
Table 5:
Distribution of cases by different grading systems.
| Characteristic | N = 6701 |
|---|---|
| Original Broders Tumor Grade | |
| 1 | 26 (4%) |
| 2 | 340 (58%) |
| 3 | 220 (38%) |
| Unknown | 84 |
| Jesinghaus Tumor Grade | |
| 1 | 25 (4%) |
| 2 | 260 (44%) |
| 3 | 302 (51%) |
| Unknown | 83 |
| Consensus Broders Tumor Grade | |
| 1 | 17 (3%) |
| 2 | 356 (61%) |
| 3 | 214 (36%) |
| Unknown | 83 |
| Silva Pattern | |
| A | 15 (3%) |
| B/C | 569 (97%) |
| Unknown | 86 |
| Broders Binary Tumor Grade | |
| High | 367 (64%) |
| Low | 204 (36%) |
| Unknown | 99 |
| Jesinghaus Binary Tumor Grade | |
| High | 385 (66%) |
| Low | 194 (34%) |
| Unknown | 91 |
n (%)
When comparing original and consensus Broders grades for evaluable cases, original Broders grade 1, 7 (35%) were consensus Broders grade 1, 12 (60%) were upgraded to consensus Broders grade 2, and 1 (5%) was upgraded to consensus Broders grade 3. Similarly, 220 (74%) of original grade 2 tumors remained consensus Broders grade 2, 7 (2%) were downgraded to consensus Broders grade 1, and 72 (24%) were upgraded to consensus Broders grade 3. Among original Broders grade 3 SCCs, 119 (62%) remained consensus grade 3, 2 (1%) were downgraded to consensus grade 1, and 72 (37%) were downgraded to consensus grade 2 (Table 6). Kappa agreement between original and consensus Broders tumor grades was low (κ=0.365, 95% CI: 0.286, 0.445).
Table 6:
Changes from original to consensus Broders tumor grade.
| Characteristic | Original Tumor Grade 1 | Original Tumor Grade 2 | Original Tumor Grade 3 |
|---|---|---|---|
|
n (%) | |||
| Consensus Tumor Grade | |||
| 1 | 7 (35%) | 7 (2%) | 2 (1%) |
| 2 | 12 (60%) | 220 (74%) | 72 (37%) |
| 3 | 1 (5%) | 72 (24%) | 119 (62%) |
Progression-free survival and overall survival status were available for 636 and 653 cases, respectively. The median follow-up among survivors was 5.7 years (range, 0–27.8 years). Progression-free survival rates were 93%, 79%, and 71%, and overall survival rates were 98%, 86%, and 79% at 1, 5, and 10 years, respectively. On univariable analysis, original Broders tumor grade, consensus Broders tumor grade, and Jesinghaus tumor grade were statistically significant for overall survival (P<0.001, P<0.034, P<0.013, respectively); only original Broders tumor grade was significant for progression-free survival (P=0.038). Silva pattern was not statistically significant for overall survival (P=0.23) or progression-free survival (P=0.10) (Tables 7, 8).
Table 7:
Univariable analysis for overall survival (OS).
| Characteristic | N | Event N | HR | 95% CI | P value |
|---|---|---|---|---|---|
| Original Broders Tumor Grade | 569 | 108 | <0.001 | ||
| 1 | — | — | |||
| 2 | 4.1 | 0.49, 35 | |||
| 3 | 5.9 | 0.77, 45 | |||
| Jesinghaus Tumor Grade | 571 | 101 | 0.013 | ||
| 1 | — | — | |||
| 2 | 1.5 | 0.48, 4.7 | |||
| 3 | 2.2 | 0.58, 8.2 | |||
| Consensus Broders Tumor Grade | 571 | 101 | 0.034 | ||
| 1 | — | — | |||
| 2 | 3.3 | 0.88, 12 | |||
| 3 | 4.2 | 1.1, 16 | |||
| Silva Pattern | 569 | 101 | 0.23 | ||
| A | — | — | |||
| B/C | 2.8 | 0.53, 14 | |||
| Broders Binary Tumor Grade | 559 | 98 | <0.001 | ||
| High | — | — | |||
| Low | 0.49 | 0.40, 0.59 | |||
| Jesinghaus Binary Tumor Grade | 564 | 100 | 0.005 | ||
| High | — | — | |||
| Low | 0.58 | 0.39, 0.85 |
HR, hazard ratio; CI, confidence interval
Table 8:
Univariable analysis for progression-free survival (PFS).
| Characteristic | N | Event N | HR | 95% CI | P value |
|---|---|---|---|---|---|
| Original Broders Tumor Grade | 554 | 143 | 0.038 | ||
| 1 | — | — | |||
| 2 | 1.4 | 0.58, 3.5 | |||
| 3 | 1.9 | 0.75, 5.0 | |||
| Jesinghaus Tumor Grade | 555 | 141 | 0.059 | ||
| 1 | — | — | |||
| 2 | 1.1 | 0.56, 2.3 | |||
| 3 | 1.5 | 0.66, 3.6 | |||
| Consensus Broders Tumor Grade | 555 | 141 | 0.076 | ||
| 1 | — | — | |||
| 2 | 4.6 | 1.0, 21 | |||
| 3 | 5.1 | 1.1, 23 | |||
| Silva Pattern | 552 | 141 | 0.10 | ||
| A | — | — | |||
| B/C | 3.9 | 0.77, 20 | |||
| Broders Binary Tumor Grade | 539 | 134 | <0.001 | ||
| High | — | — | |||
| Low | 0.63 | 0.51, 0.78 | |||
| Jesinghaus Binary Tumor Grade | 547 | 138 | 0.057 | ||
| High | — | — | |||
| Low | 0.70 | 0.48, 1.0 |
HR, hazard ratio; CI, confidence interval.
Predictive accuracy for overall survival was 0.559 (original Broders tumor grade), 0.542 (consensus Broders tumor grade), 0.554 (Jesinghaus tumor grade), and 0.512 (Silva pattern). Predictive accuracy for progression-free survival was 0.542 (original Broders tumor grade), 0.525 (consensus Broders tumor grade), 0.541 (Jesinghaus tumor grade), and 0.515 (Silva pattern) (Tables 9, 10).
Table 9:
Predictive accuracy of each tumor grade, overall survival.
| Variable | CPE | Lower 95% CI | Upper 95% CI |
|---|---|---|---|
| Original Broders Tumor Grade | 0.559 | 0.514 | 0.604 |
| Consensus Broders Tumor Grade | 0.542 | 0.496 | 0.588 |
| Jesinghaus Tumor Grade | 0.554 | 0.503 | 0.605 |
| Silva Pattern | 0.512 | 0.491 | 0.533 |
| Broders Binary Tumor Grade | 0.579 | 0.53 | 0.628 |
| Jesinghaus Binary Tumor Grade | 0.56 | 0.511 | 0.609 |
CPE, concordance probability estimate; CI, confidence interval.
Table 10:
Predictive accuracy of each tumor grade, progression-free survival.
| Variable | CPE | Lower 95% CI | Upper 95% CI |
|---|---|---|---|
| Original Broders Tumor Grade | 0.542 | 0.502 | 0.582 |
| Consensus Broders Tumor Grade | 0.525 | 0.485 | 0.565 |
| Jesinghaus Tumor Grade | 0.541 | 0.498 | 0.584 |
| Silva Pattern | 0.515 | 0.498 | 0.532 |
| Broders Binary Tumor Grade | 0.552 | 0.51 | 0.594 |
| Jesinghaus Binary Tumor Grade | 0.54 | 0.499 | 0.581 |
CPE, concordance probability estimate; CI, confidence interval.
Associations with the surgical margin status were affected by changes from the original to consensus Broders grade (P<0.001) (Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/IJGP/A154) and associations with stage, greatest tumor size, depth of invasion, and tumor stroma were affected by changes from the consensus Broders to Jesinghaus tumor grade (Supplemental Table 2, Supplemental Digital Content 1, http://links.lww.com/IJGP/A154).
For the proposed binary grading system (Table 5, Figure 4), 64% of cases fit criteria for Broders binary high grade and 66% for Jesinghaus binary high grade. Both Broders binary grade and Jesinghaus binary grade were statistically significant in univariable analysis for overall survival; however, only Broders binary grade was significant for progression-free survival (Tables 7, 8; Figures 6, 7). Predictive accuracy indices for overall survival and profression free survival were improved in both the Broders and Jesinghaus binary tumor grading systems compared to grading systems that do not account for tumor stroma (Tables 9, 10).
Figure 6:
Kaplan-Meier plots by grading systems for overall survival (OS). (A) Original Broders tumor grade; (B) Jesinghaus tumor grade; (C) Consensus Broders tumor grade; (D) Silva pattern; (E) Broders binary tumor grade; and (F) Jesinghaus binary tumor grade.
Figure 7:
Kaplan-Meier plots by grading system for progression-free survival (PFS). (A) Original Broders tumor grade; (B) Jesinghaus tumor grade; (C) Consensus Broders tumor grade; (D) Silva pattern; (E) Broders binary tumor grade; and (F) Jesinghaus binary tumor grade.
Logistic regression for presence of lymph node metastases showed that Jesinghaus tumor grade (P<0.001) as well as both binary systems (Broders grade, P=0.007; Jesinghaus grade, P<0.001) were associated with presence of lymph node metastases (Supplemental Table 3, Supplemental Digital Content 1, http://links.lww.com/IJGP/A154).
Discussion
To our knowledge, this is the largest study to analyze various grading systems for SCC. Our findings demonstrate that neither Broders classification nor the Jesinghaus or Silva tumor grading systems have high predictive acccuracy for overall survival or progression-free survival. Incorporating tumor stroma morphology into the Broders classification and Jesinghaus grading systems as a binary algorithm better predicted survival and presence of lymph node metastases, which may impact surgical management of patients with SCCs. Kappa agreement between the grading systems was low, which suggests that despite rigid definitions and thresholds, grading SCCs without accounting for tumor stroma is subjective and may be of limited value.
Recent international classifications and guidelines do not recommend grading SCCs.1,10 While most clinicians expect tumor grade to be included in the pathology report, whether or not it will guide treatment strategy, there is a lack of consensus in the literature as to the appropriate uses and prognostic values of current grading systems. Most studies on prognostic parameters including histologic grade analyzed SCCs together with endocervical adenocarcinomas and adenosquamous carcinomas even though they are different tumor types. These tumors are also currently graded using different methods.10,19 Futhermore, some published studies do not provide detail about the criteria for assigning grade.20,21 Broders histologic classification system and modifications thereof have been used by the Gynecologic Oncology Group (GOG) for decades and frequently used worldwide. This grading system has limitations, though, such as subjectivity in scoring the degree of keratinization and nuclear atypia. Consequently, Broders classification system has not gained universal acceptance.
In a study of 188 patients with FIGO stage IIA or B SCC, Stendhal et al proposed a multifactorial grading system in the form of a histopathologic malignancy index that included 8 features: growth pattern, differentiation, nuclear pleomorphism, mitotic figures, pattern of invasion, depth of invasion, lymphovascular invasion, and inflammatory reaction.22,23 Their system, however, has not been adopted in routine practice. Neither keratinization nor pleomorphism was correlated with survival, while not surprisngly, lymphovascular invasion carried the strongest prognostic value.23
Zaino et al compared different grading systems (Broders grade, nuclear pleomorphism, mitotic rate and density of lymphoid, plasma cell, and eosinophil infiltrates into the tumor, and pattern of tumor invasion at the stromal interface) using 195 FIGO stage IB SCCs and found that none of the existing systems predicted presence of lymph node metastases and progression-free interval.6 More recent data suggest that pattern of invasion, similar to those designed for endocervical adenocarcinomas, could have clinical relevance; however, these findings have not been confirmed in a large international cohort of SCCs.13,16 Solid tumor cords (closed/pushing and finger-like pattern) have been reported to have a better prognosis than a spray-like pattern of invasion, which is associated with advanced stage, pelvic lymph node metastases, and reduced survival. Interestingly, the significance persists even when this feature is discerned only in a small biopsy.13,14
Jesinghaus et al proposed a grading system based on the number of buds and the size of buds evaluated in the entire tumor, which had significant associations with overall survival, disease-specific survival, and disease-free-survival on test and validation cohorts (125 and 122 cases, respectively), using resection specimens only.3 This grading system was validated in one subsequent study, which analyzed an institutional cohort of 94 consecutive cases of stage pT1b or higher disease. The full scoring system was associated with decreased overall survival, disease-specific survival, and disease-free survival, independent of patient age, pathologic stage, and regional lymph node status.24
In our study, two pathologists reviewed all cases and assigned the Broders grade and compared this to the originally assigned grade, finding many grading differences between the two. Despite this, the original and consensus Broders grades correlated with overall survival on univariable analysis, and, interestingly, the original Broders grade also correlated with progression-free survival. The concordance probability estimates, however, were very low, which suggests neither of these grading systems can accurately predict overall survival and progression free, and supports the exclusion of Broders grade from classifications and management guidelines.
Invasion by malignant tumor cells into surrounding native tissue induces a new matrix, the tumor stroma. In SCCs from a variety of sites, spray-like morphology (small groups of infiltrating cells and high tumor cell dissociation) with a myxoid/desmoplastic tumor stroma can be associated with worse survival.14 Consequently, a myxoid tumor stroma may be an indicator of poor survival in cervical SCC, as has been described even in endometrial endometrioid adenocarcinoma.25,26 Our study demonstrated that integrating tumor stroma morphology into the Broders classification and Jesinghaus grading systems improved their prognostic performances. Since many pathologists still use the Broders classification system, despite concerns of reproducibility, this system can be applied even to a biopsy and used in combination with myxoid stroma to predict survival. While the Jesinghaus grading system is more complex and restricted to resection specimens only, the parameters are more objective.
Logistic regression for presence of lymph node metastases showed that both binary grading systems were associated with the presence of lymph node metastases, which may impact surgical management, including patient selection for lymph node dissection.
This study is not without limitations. Despite the inclusion of 10 international centers, this study is limited by the retrospective nature of case collection, chart review, and pathology review. Also, because some of the original pathology slides were missing, not all cases were re-reviewed for histologic grade.
Histologic grade has poor reproducibility and limited predictive accuracy for SCC. Incorporating tumor stroma morphology into the Broders classification and Jesinghaus grading systems expressed as low- or high-grade better predicted survival and the presence of lymph node metastases when compared to existing schemes.
Supplementary Material
Funding:
This work was supported in part by a Cancer Center Support Grant by the NIH/NCI (P30 CA008748).
Footnotes
Declarations
Conflicts of interest/competing interests: Outside the submitted work, N.R. Abu-Rustum reports research funding paid to the institution from GRAIL. A. Iasonos reports consulting fees from Mylan. All other authors have no potential conflicts of interest to disclose.
Availability of data and material: All data can be provided by the corresponding author (Simona Stolnicu).
Ethics approval: This study was approved by the Institutional Review Boards of each participating institution. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee, and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.
Consent to participate: Informed consent was obtained from all individual participants included in the study. In cohorts without direct informed consent, contributing institutions obtained an approved waiver of consent from the local institutional research ethics board allowing use of the study material.
Consent for publication: All authors agreed to publish the results of the present study.
References
- 1.Herrington CS; WHO Classification of Tumours Editorial Board. WHO Classification of Tumours Female Genital Tumours 5th ed. International Agency for Research on Cancer; 2020. [Google Scholar]
- 2.Broders AC. Squamous cell cancer of the lip: a study of five hundred and thirty-seven cases. JAMA 1920;74(10):656–64. [Google Scholar]
- 3.Jesinghaus M, Strehl J, Boxberg M, et al. Introducing a novel highly prognostic grading scheme based on tumor budding and cell nest size for squamous cell carcinoma of the uterine cervix. J Pathol Clin Res 2018;4(2):93–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Jesinghaus M, Boxberg M, Konukiewitz B, et al. A novel grading system based on tumor budding and cell nest size is a strong predictor of patient outcome in esophageal squamous cell carcinoma. Am J Surg Pathol 2017;41(8):1112–20. [DOI] [PubMed] [Google Scholar]
- 5.Benda JA. Histopathologic prognostic factors in early stage cervical carcinoma. J Natl Cancer Inst Monogr 1996;(21):27–34. [PubMed] [Google Scholar]
- 6.Zaino R, Ward S, Delgado G, et al. Histopathologic predictors of the behavior of surgically treated stage IB squamous cell carcinoma of the cervix. A gynecologic oncology group study. Cancer 1992;69(7):1750–8. [DOI] [PubMed] [Google Scholar]
- 7.Tiltman AJ. The pathology of cervical tumours. Best Pract Res Clin Obstet Gynaecol 2005;19(4):485–500. [DOI] [PubMed] [Google Scholar]
- 8.Cibula D, Potter R, Planchamp F, et al. The European Society of Gynaecological Oncology/European Society for Radiotherapy and Oncology/European Society of Pathology guidelines for the management of patients with cervical cancer. Int J Gynecol Cancer 2018;28(4):641–55. [DOI] [PubMed] [Google Scholar]
- 9.McCluggage WG. Towards developing a meaningful grading system for cervical squamous cell carcinoma. J Pathol Clin Res 2018;4(2):81–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Park KJ, Selinger CI, Alvarado-Cabrero I, et al. Dataset for the reporting of carcinoma of the cervix: recommendations from the International Collaboration on Cancer Reporting (ICCR). Int J Gynecol Pathol 2022;41(Suppl 1):S64–S89. [DOI] [PubMed] [Google Scholar]
- 11.Ambros RA, Malfetano JH, Mihm MC Jr. Clinicopathologic features of vulvar squamous cell carcinomas exhibiting prominent fibromyxoid stromal response. Int J Gynecol Pathol 1996;15(2):137–45. [DOI] [PubMed] [Google Scholar]
- 12.Murray SK, Young RH, Scully ER. Unusual epithelial and stromal changes in myoinvasive endometrioid adenocarcinoma: a study of their frequency, associated diagnostic problems and prognostic significance. Int J Gynecol Pathol 2003;22(4):324–33. [DOI] [PubMed] [Google Scholar]
- 13.Horn LC, Fischer U, Raptis G, et al. Pattern of invasion is of prognostic value in surgically treated cervical cancer patients. Gynecol Oncol 2006;103(3):906–11. [DOI] [PubMed] [Google Scholar]
- 14.Horn LC, Hommel N, Roschlau U, et al. Peritumoral stroma remodeling, pattern of invasion and expression of c-met/HGF in advanced squamous cell carcinoma of the cervix uteri, FIGO stages III and IV. Eur J Obstet Gynecol Reprod Biol 2012;163(1):76–80. [DOI] [PubMed] [Google Scholar]
- 15.Bichel P, Jakobsen A. Histopathologic grading and prognosis of uterine cervical carcinoma. Am J Clin Oncol 1985;8(3):247–54. [DOI] [PubMed] [Google Scholar]
- 16.Diaz De Vivar A, Roma AA, Park KJ, et al. Invasive endocervical adenocarcinoma: proposal for a new pattern-based classification system with significant clinical implications: a multi-institutional study. Int J Gynecol Pathol 2013;32(6):592–601. [DOI] [PubMed] [Google Scholar]
- 17.Mo Q, Gonen M, Heller G. CPE: Concordance Probability Estimates in Survival Analysis. R package version 1.5.1; 2018. https://CRAN.R-project.org/package=CPE
- 18.R Core Team. R: A language and environment for statistical computing R Foundation for Statistical Computing; 2021. [Google Scholar]
- 19.Talia KL, Oliva E, Rabban JT, et al. Grading of endocervical adenocarcinomas: review of the literature and recommendations from the International Society of Gynecological Pathologists. Int J Gynecol Pathol 2021;40(Suppl 1):S66–S74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Vinh-Hung V, Bourgain C, Vlastos G, et al. Prognostic value of histopathology and trends in cervical cancer: a SEER population study. BMC Cancer 2007;7:164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Macdonald OK, Chen J, Dodson M, et al. Prognostic significance of histology and positive lymph node involvement following radical hysterectomy in carcinoma of the cervix. Am J Clin Oncol 2009;32(4):411–6. [DOI] [PubMed] [Google Scholar]
- 22.Stendahl U, Willen H, Willen R. Classification and grading of invasive squamous cell carcinoma of the uterine cervix. Acta Radiol Oncol Radiat Phys Biol 1979;18(6):481–96. [DOI] [PubMed] [Google Scholar]
- 23.Stendahl U, Eklund G, Willen R. Invasive squamous cell carcinoma of the uterine cervix. IV. Analysis of a histopathologic malignancy grading system and construction of a partial index. Acta Radiol Oncol 1981;20(5):289–94. [DOI] [PubMed] [Google Scholar]
- 24.Zare SY, Aisagbonhi A, Hasteh F, et al. Independent validation of tumor budding activity and cell nest size as determinants of patient outcome in squamous cell carcinoa of the uterine cervix. Am J Surg Pathol 2020;44(9):1151–60. [DOI] [PubMed] [Google Scholar]
- 25.Breuninger H, Schaumburg-Lever G, Holzschuh J, Horny HP. Desmoplastic squamous cell carcinoma of skin and vermilion surface: a highly malignant subtype of skin cancer. Cancer 1997;79(5):915–9. [DOI] [PubMed] [Google Scholar]
- 26.Spiro RH, Guillamondegui O Jr, Paulino AF, Huvos AG. Pattern of invasion and margin assessment in patients with oral tongue cancer. Head Neck 1999;21(5):408–13. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.