Clinical features of anti-HMGCR-positive IMNM patients
We identified 18 patients diagnosed with anti-HMGCR-positive IMNM. Clinical data are presented in Table 1.
Table 1
Clinical features and pathological manifestations of 18 patients with anti-HMGCR-positive IMNM
| Gender | Onset age (y) | Duration (m) | Statins exposure | Muscular symptoms | Extra-muscular symptoms |
Patient 1 | Female | 24 | 4 | None | Muscle weakness, myalgia | None |
Patient 2 | Female | 24 | 2 | None | Muscle weakness | None |
Patient 3 | Female | 42 | 13 | None | Muscle weakness | None |
Patient 4 | Male | 54 | 9 | None | Muscle weakness | None |
Patient 5 | Male | 51 | 132 | None | Muscle weakness | None |
Patient 6 | Female | 63 | 1 | Rosuvastatin | Muscle weakness | None |
Patient 7 | Male | 21 | 14 | None | Muscle weakness | Rash |
Patient 8 | Female | 52 | 11 | None | Muscle weakness | None |
Patient 9 | Male | 12 | 8 | None | Muscle weakness, myalgia, dysphagia | None |
Patient 10 | Female | 70 | 1 | None | Muscle weakness | Arthralgia, cardiac involvement, weight loss |
Patient 11 | Female | 73 | 1 | None | Dyspnea | ILD, weight loss |
Patient 12 | Male | 50 | 72 | None | Muscle weakness, myalgia | Rash |
Patient 13 | Female | 43 | 2 | None | Muscle weakness, myalgia, dysphagia | Arthralgia, rash, ILD, weight loss |
Patient 14 | Female | 30 | 12 | None | Myalgia | Arthralgia |
Patient 15 | Female | 37 | 4 | None | Muscle weakness, dysphagia | None |
Patient 16 | Male | 45 | 10 | None | Muscle weakness, myalgia | Arthralgia, weight loss |
Patient 17 | Female | 21 | 84 | None | Muscle weakness | None |
Patient 18 | Male | 19 | 20 | None | Muscle weakness | None |
Table 1
| MMT-8 score before/after therapy | Highest CK(IU/L) before/after therapy | Anti-HMGCR antibody titer | Other MSAs | Treatment | Prognosis |
Patient 1 | 61/63 | 11914/2856 | +++ | None | GC | Poor |
Patient 2 | 62/69 | 7202/1453 | +++ | None | GC, IS, IVIg | Good |
Patient 3 | 66/70 | 6865/2735 | +++ | None | GC, IS | Poor |
Patient 4 | 70/73 | 3695/2534 | +++ | None | GC, IS, IVIg | Poor |
Patient 5 | 60/60 | 1164/195 | +++ | None | GC, IS | Poor |
Patient 6 | 64/72 | 4974/3113 | +++ | None | GC, IS, IVIg | Good |
Patient 7 | 71/76 | 9708/1594 | +++ | None | GC, IS, IVIg | Poor |
Patient 8 | 71/77 | 3759/1105 | +++ | None | GC, IS, IVIg | Poor |
Patient 9 | 68/75 | 589/199 | + | anti-NXP2, anti-Ku, anti-PM/Scl75 | GC | Good |
Patient 10 | 65/73 | 12555/1880 | +++ | anti-SRP, anti-SAE1 | GC, IS | Good |
Patient 11 | 80/80 | 36/15 | ++ | anti-Ro52, anti-PM/Scl75 | GC, IS | Good |
Patient 12 | 75/80 | 5215/620 | + | anti-SRP, anti-OJ, anti-Ro-52, ANA | GC, IS | Good |
Patient 13 | 64/80 | 589/126 | + | anti-MDA5, anti-PL7, anti-Ro52 | GC, IS | Good |
Patient 14 | 80/80 | 862/519 | ++ | anti-OJ, anti-Ku | GC, IS | Good |
Patient 15 | 58/65 | 23462/3532 | + | anti-SRP, anti-Ro-52, ANA | GC, IS, IVIg | Poor |
Patient 16 | 69/74 | 16037/1320 | ++ | anti-SRP | GC, IS | Good |
Patient 17 | 49/49 | 2106/153 | +++ | None | GC, IS, IVIg | Poor |
Patient 18 | 54/61 | 21410/5253 | ++ | None | GC, IS | Poor |
ANA anti-nuclear antibody, CK creatine kinase, GC glucocorticoid, IS immunosuppressant, IVIg intravenous immunoglobulin, MMT-8 manual muscle, MSA myositis-specific antibodies
Table 1 continued
| Myofiber type pathology | Myofiber type pathology | Myofiber type pathology |
| Atrophy | Necrosis | Regeneration | Heterogeneity | Central nucleus | Features | Inflammatory cell | Sarcolemma/ Capillary MAC | MHC-Ⅰ |
Patient 1 | 15% | 15% | 15% | 45% | 1% | endomysial fibrosis | endomysium | +/- | Focal |
Patient 2 | 13% | 8% | 10% | 31% | 2% | None | endomysium | +/- | Focal |
Patient 3 | 10% | 9% | 3% | 22% | 1% | None | None | +/- | Focal |
Patient 4 | 18% | 19% | 6% | 43% | 9% | None | endomysium | +/- | Focal |
Patient 5 | NA | NA | NA | NA | NA | NA | NA | NA | NA |
Patient 6 | 14% | 3% | 7% | 24% | 3% | None | endomysium | +/- | Focal |
Patient 7 | 8% | 2% | 5% | 15% | 22% | None | endomysium | +/- | Focal |
Patient 8 | 9% | 9% | 9% | 27% | 3% | None | endomysium | +/- | Focal |
Patient 9 | 8% | 3% | 3% | 14% | 1% | perimysial atrophy | perimysium | +/ + | Diffuse |
Patient 10 | 5% | 3% | 15% | 23% | 2% | None | None | +/- | Focal |
Patient 11 | 10% | 0% | 1% | 11% | 1% | None | None | +/ + | Focal |
Patient 12 | NA | NA | NA | NA | NA | NA | NA | NA | NA |
Patient 13 | NA | NA | NA | NA | NA | NA | NA | NA | NA |
Patient 14 | 5% | 1% | 3% | 9% | 2% | vasculitis | perimysium, endomysium | +/- | Focal |
Patient 15 | 3% | 12% | 9% | 24% | 1% | None | endomysium | +/- | Focal |
Patient 16 | 2% | 10% | 12% | 24% | 1% | None | perimysium, endomysium | +/- | Focal |
Patient 17 | 16% | 6% | 7% | 29% | 22% | endomysial fibrosis | perimysium, endomysium | +/- | Focal |
Patient 18 | 34% | 4% | 6% | 44% | 20% | endomysial fibrosis | endomysium | +/- | Focal |
MAC membrane attack complex, MHC-Ⅰ major histocompatibility complex Ⅰ, NA not available.
The age of onset for these patients ranged from 19 to 73 years of age (40.61 ± 18.28 years). Eleven patients (11/18, 61.11%) were female and seven patients (7/18, 38.89%) were male. The median duration from symptom emergence to diagnosis for anti-HMGCR-positive IMNM was 9.50 months (1month-132 months).
Seventeen patients (17/18, 94.44%) complained of muscular symptoms, sixteen patients (16/18, 88.89%) had muscle weakness, five patients (5/18, 27.78%) had myalgia, and two patients (2/18, 11.1%) had dyspnea and two patients (2/18, 11.11%) had dysphagia. Seven patients (7/18, 38.9%) had extra-muscular symptoms. Among them, four patients (4/18, 22.22%) had arthralgia and four patients (4/18, 22.22%) had weight loss, three patients (3/18, 16.67%) had a rash, two patients (2/18, 11.11%) had interstitial lung disease (ILD), and one patient (1/18, 5.56%) had cardiac involvement. Nasopharyngeal carcinoma was found in one patient (1/18, 5.56%) during outpatient follow-up. One patient (1/18, 5.56%) had an exposure history to rosuvastatin.
All patients were evaluated for muscle strength based on the MMT-8. The score for theMMT-8 ranged from 49 to 80 (65.94 ± 8.17). CK levels were elevated in 17 patients (17/18, 94.44%), while one patient (1/18, 5.56%) remained normal. The serum CK level ranged from 36 to 23462 IU/L (7341.22 ± 7177.24 IU/L).
Anti-HMGCR was strongly positive (+++) for ten patients (10/18, 55.56%), moderately positive (++) for four patients (4/18, 22.22%) and weakly positive (+) for four patients (4/18, 22.22%). Anti-HMGCR antibody overlap with other antibodies was observed in eight patients (8/18, 44.44%). The coexistence of anti-HMGCR antibody with one or more MSA and/or MAA was detected in all eight of these patients (8/18, 44.44%), including anti-SAE1 antibody (1/8, 12.50%), anti-MDA5 antibody (1/8, 12.50%), anti-NXP2 antibody (1/8, 12.50%), anti-PL-7 antibody (1/8, 12.50%), anti-Ku antibody (2/8, 25.00%), anti-OJ antibody (2/8, 25.00%), anti-PM-Scl-75 antibody (2/8, 25.00%), anti-Ro-52antibody (4/8, 50.00%), and anti-SRP antibody (4/8, 50.00%). ANA were detected in two patients (2/8, 25.00%).
Pathologic manifestations of anti-HMGCR-positive IMNM patients
Muscle biopsy results for 15 patients are presented in Table 1. As for myo-fiber type pathology, the percentage of myo-fibers with atrophy, necrosis, regeneration, heterogeneity, and central nucleus was a range of 2%-34%, 0–19%, 1%-15%, 9–45%, and 1%-22%, respectively.
Regarding muscle tissue pathology, three patients (3/15, 20.00%) had endomysial fibrosis, three patients (3/15, 20.00%) had vasculitis, one patient (1/15, 6.67%) had perimysial atrophy.
With respect to inflammation, eleven patients (11/15, 73.33%) had endomysial inflammatory cell infiltration, while four patients (4/15, 26.67%) had perimysial inflammatory cell infiltration. Fourteen patients (14/15, 93.33%) had focal distribution of MHC-Ⅰ, while one patient (1/15, 6.67%) had diffuse distribution of MHC-Ⅰ. Sarcolemma MAC deposition was observed in all muscle biopsies (15/15, 100.00%), while capillary MAC positivity was only observed in two patients (2/15, 13.33%). Immunohistochemistry showed that CD4+ T cells, CD8+ T cells, and CD68+ macrophages were the most common inflammatory cells. No diagnostic mutations of known LGMD associated proteins were observed by immunostaining.
Treatment and response of anti-HMGCR-positive IMNM patients
All 18 patients received treatment (Table 1). Nine patients (9/18, 50.00%) were treated with dual-therapy of GC and IS. Seven patients (7/18, 38.89%) received triple-therapy of GC, IS, and IVIg. Two patients (2/18, 11.11%) received mono-therapy with GC. After treatment, myalgia and extra-muscular symptoms disappeared in all patients. All patients had a reduction in CK. Serum CK levels after treatment ranged from 15 to 5253 IU/L (1622.33 ± 1462.54 IU/L), with a reduction range of 31.42%-92.74% (71.29%±18.86%). Except for two patients with normal muscle strength, 14 patients (14/18, 77.8%) had variable muscle strength development, while two patients (2/18, 11.11%) exhibited stabilization. MMT-8 scores after treatment varied from 49 to 80 (70.94 ± 8.50), with an improvement variation of 0–16 (5.00 ± 3.99).
Comparison of non-overlap and overlap anti-HMGCR-positive IMNM patients
We collected clinical data for 50 patients diagnosed with anti-HMGCR-positive IMNM who had been reported in previous studies as well as 18 patients from our hospital.
Since the presence of anti-HMGCR antibody and other antibodies resulted in different clinical features, we separated patients into two subgroups: non-overlap anti-HMGCR-positive IMNM patients and overlap anti-HMGCR-positive IMNM patients.
The comparison of clinical data for these two subgroups is presented in Table 2. Compared to overlap anti-HMGCR-positive IMNM patients, earlier onset age (36.65 ± 18.70 vs 46.14 ± 17.48, p = 0.038) and longer disease duration (56.25 ± 80.58 vs11.53 ± 19.19, p = 0.015) were found in non-overlap patients. Moreover, non-overlap anti-HMGCR patients had a higher frequency of muscle weakness (100.00% vs 75.00%, p = 0.001), but a lower prevalence of myalgia (17.50% vs 46.43%, p = 0.010) and dyspnea (0.00% vs 25.00%, p = 0.001), with lower MMT-8 scores (63.44 ± 7.21 vs 71.77 ± 7.14, p = 0.004) than overlap patients. With regard to extra-muscular symptoms, overlap anti-HMGCR-positive IMNM patients had a higher prevalence of extra-muscular symptoms (89.29% vs 17.50%, p < 0.0001), including ILD (28.57% vs 2.50%, p = 0.003), arthralgia (32.10% vs 0.00%, p = 0.0001), weight (25.00% vs 2.50%, p = 0.014), and skin involvement (53.57% vs 15.00%, p = 0.001) than non-overlap patients. Overlap anti-HMGCR-positive IMNM patients had a lower frequency of treatment with IVIg (9.50% vs 60.00%, p = 0.001) than non-overlap patients. Correlations between overlap antibodies and muscular and extra-muscular symptoms in overlap anti-HMGCR-positive IMNM patients are shown in Fig. 1. With regard to muscular symptoms, limb muscle weakness was predominantly presented in patients except for patients with anti-MDA5 antibody, all of which (5/5, 100.0%) complained of dyspnea, and patients with anti-PM-Scl-75 antibody presenting with dysphagia (2/3, 66.67%). Myalgia was shown in all patients with anti-Ku antibody (3/3, 100.00%) and anti-OJ antibody (2/2, 100.00%). As for extra-muscular symptoms, all patients with anti-MDA5 antibody (5/5, 100.00%) showed skin involvement and ILD. Most patients with anti-Jo-1 antibody (2/3, 66.67%) presented with ILD and arthralgia.
Table 2
Comparison of clinical features between non-overlap and overlap patients
| Non-overlap (n = 40) | Overlap (n = 28) | p |
Clinical data | | | |
Female | 23/36 (63.89) | 16 (57.14) | 0.583 |
Onset age (years old) | 36.65 ± 18.70 | 46.14 ± 17.48 | 0.038* |
Duration (months) | 56.25 ± 80.58 | 11.53 ± 19.19 | 0.015 |
MMT-8 | 63.44 ± 7.21 | 71.77 ± 7.14 | 0.004* |
Highest CK (IU/L) | 7401.54 ± 5616.61 | 8470.60 ± 7289.30 | 0.609 |
Clinical symptoms | | | |
Muscular symptoms | 40 (100.00) | 28 (100.00) | 1.000 |
Muscle weakness | 40 (100.00) | 21 (75.00) | 0.001* |
Myalgia | 7 (17.50) | 13 (46.43) | 0.010* |
Dyspnea | 0 (0.00) | 7 (25.00) | 0.001* |
Dysphagia | 5 (12.50) | 8 (28.57) | 0.097 |
Extra-muscular symptoms | 7 (17.50) | 25 (89.29) | < 0.0001* |
ILD | 1 (2.50) | 8 (28.57) | 0.003* |
Arthralgia | 0 (0.00) | 9 (32.14) | 0.0001* |
Loss of weight | 1 (2.50) | 7 (25.00) | 0.014* |
Skin involvement | 6 (15.00) | 15 (53.57) | 0.001* |
Treatment and outcomes | |
With IVIg | 21/35 (60.00) | 2/21 (9.50) | 0.001* |
Good prognosis | 2/10 (20.00) | 7/8 (87.50) | 0.015* |
CK creatine kinase, ILD interstitial lung disease, IVIg intravenous immunoglobulin, MMT-8 manual muscle testing. * P < 0.05.
The muscular and extra-muscular symptoms in all overlap anti-HMGCR-positive IMNM patients. The heatmap showed the percentage of muscular and extra-muscular symptoms in patients with different overlap antibodies. With regard to muscular symptoms, limb muscle weakness was predominantly presented in patients except for patients Anti-MDA5 antibody complaining of with dyspnea (5/5, 100.00%) and anti-PM-Scl-75 antibody of dysphagia (2/3, 66.67%). Myalgia was shown in all patients with anti-Ku antibody (3/3, 100.00%) and anti-OJ antibody (2/2, 100.00%). As for extra-muscular symptoms, all patients with anti-MDA5 antibody (5/5, 100.00%) showed skin involvement and ILD. Most patients with anti-Jo-1 antibody (2/3, 66.67%) presented with ILD and arthralgia.
Comparison of the 15 biopsies from non-overlap and overlap anti-HMGCR-positive IMNM patients from our hospital is shown (Fig. 2). As for myo-fiber type pathology, non-overlap anti-HMGCR-positive IMNM patients had a higher percentage of myo-fibers with atrophy (15.22%±7.79% vs 5.50%±3.02%, p = 0.013), central nucleus (9.22%±9.40% vs 1.33%±0.52%, p = 0.028) and heterogeneity (31.11%±10.69% vs 17.50%±6.95%, p = 0.017) compared to overlap patients. No statistically significant difference was observed in myo-fibers with necrosis (8.33%±5.61% vs 4.83%±4.96%, p = 0.238) and regeneration (7.56%±3.47% vs 7.17%±5.67%, p = 0.871). For muscle tissue pathology, we found no significant difference in the prevalence of perimysial atrophy (0.00% vs 16.67%, p = 0.400), endomysial fibrosis (33.33% vs 0.00%, p = 0.229), or vasculitis (11.11% vs 16.67%, p = 1.000) between overlap and non-overlap anti-HMGCR-positive IMNM patients. With regard to inflammatory pathology, there was no difference in endomysial inflammatory cell infiltration (88.89% vs 50.00%, p = 0.235), perimysial inflammatory cells infiltration (11.11% vs 50.00%, p = 0.235), the focal distribution of MHC-Ⅰ (100.00% vs 83.33%, p = 0.400), sarcolemma MAC deposit (100.00% vs 100.00%, p = 1.000), or capillary MAC deposit (0.00% vs 33.33%, p = 0.143) between non-overlap and overlap anti-HMGCR-positive IMNM patients.
Comparison of non-overlap and overlap anti-HMGCR-positive IMNM patient pathologic manifestations. For myo-fiber pathology, non-overlap anti-HMGCR-positive IMNM presented with a higher percentage of myo-fibers with atrophy (15.22%±7.79% vs 5.50%±3.02%, p = 0.013), central nucleus (9.22%±9.40% vs 1.33%±0.52%, p = 0.028) and heterogeneity (31.11%±10.69% vs 17.50%±6.95%, p = 0.017) compared to overlap patients. For muscle tissue pathology and inflammatory pathology, there were no significant differences.
Differences in treatment between non-overlap and overlap anti-HMGCR-positive IMNM patients are shown in Table 2. Non-overlap anti-HMGCR-positive IMNM patients had a higher frequency of IVIg use than overlap patients (60.00% vs 9.50%, p = 0.001).
Response to the treatment is presented in Fig. 3A, B. Compared with non-overlap anti-HMGCR-positive IMNM patients, overlap patients had greater improvement (8.00 ± 4.10 vs 4.20 ± 2.90, p = 0.047) in MMT-8 scores after treatment. However, there was no significant difference in the percentage of CK reduction (69.05%±20.22% vs 74.10%±17.94%, p = 0.588) between non-overlap and overlap anti-HMGCR-positive IMNM patients. Additionally, overlap anti-HMGCR-positive IMNM patients had a higher prevalence of a good prognosis than non-overlap patients (87.50% vs 20.00%, p = 0.015) (Table 2).
A comparison of patients with a good and a poor prognosis was performed. Anti-HMGCR-positive IMNM patients who had a good prognosis had higher MMT-8 scores compared to patients who had a poor prognosis (69.67 ± 6.98 vs 62.22 ± 7.87, p = 0.050) (Suppl. Table 1). By univariable analysis, overlap antibody (B = 28.0, p = 0.012) was found to be a good prognostic predictor, while MMT-8 score was not (B = 1.2, p = 0.079). By multivariable analysis, overlap antibody (B = 28.0, p = 0.012) was identified as an independent predictor of a good prognosis (Suppl. Table 2). ROC analysis demonstrated a good performance for overlap antibody as a predictor of prognosis, with an AUC of 0.833 (p = 0.017, 95%CI, 0.629-1.000) (Fig. 3C). A cut-off value (1.50) had a sensitivity of 77.8% and specificity of 88.9%.
Treatment response and prognosis for non-overlap and overlap anti-HMGCR-positive IMNM patients. (A) Percentage of CK reduction: there was no significant difference between non-overlap and overlap patients. (B) MMT-8 improvement: overlap patients had greater improvement in MMT-8 scores after treatment (8.00 ± 4.10 vs 4.20 ± 2.90, p = 0.047) than non-overlap patients. (C) The area under curve (AUC) values indicate that overlap antibody can serve as a predictor of a good prognosis, with an AUC of 0.833 (p = 0.017, 95%CI, 0.629-1.000). A cut-off value (1.50) had a sensitivity of 77.8% and specificity of 88.9%. Error bar = Standard deviation. *p < 0.05.
Comparison of non-LGMD-like and LGMD-like anti-HMGCR-positive IMNM patients
Among the patients diagnosed with non-overlap anti-HMGCR-positive IMNM, a special type of disease resembling LGMD has been identified that does not have an associated pathogenic gene. To demonstrate whether the LGMD-like anti-HMGCR-positive IMNM is a unique subtype of non-overlap anti-HMGCR-positive IMNM, we separated non-overlap anti-HMGCR-positive IMNM patients into two phenotypes: non-LGMD-like anti-HMNGCR-positive IMNM and LGMD-like anti-HMNGCR-positive IMNM. Comparisons were made of the clinical-pathologic characteristics of the two phenotypes (Table 3).
With regard to clinical characteristics, LGMD-like anti-HMGCR-positive IMNM patients had an earlier age of onset (21.33 ± 14.13 vs 44.68 ± 16.94, p < 0.0001), longer duration (128.75 ± 97.92 vs 15.90 ± 28.80, p < 0.0001), and lower MMT-8 scores (51.50 ± 3.54 vs 68.33 ± 7.21 p = 0.003) than non-LGMD-like patients.
With regard to myo-fiber type pathology, LGMD-like anti-HMGCR-positive patients exhibited profound myo-fibers with atrophy and predominant myo-fibers with the central nucleus, resembling muscular dystrophy more than the shared characteristics of severe myo-fiber necrosis and regeneration (Fig. 4). There was no remarkable difference with regard to muscle tissue or inflammatory pathology. With regard to different types of CD staining for T lymphocytic cell, there were also no significant differences.
Table 3
Comparison of clinical features between non-LGMD-like and LGMD-like patients
| Non-LGMD-like (n = 28) | LGMD-like (n = 12) | p |
Clinical data | | | |
Female | 18 (64.29) | 4/8 (50.00) | 0.683 |
Onset age (years old) | 44.68 ± 16.94 | 21.33 ± 14.13 | < 0.0001* |
Duration (months) | 15.90 ± 28.80 | 128.75 ± 97.92 | < 0.0001* |
MMT-8 | 68.33 ± 7.21 | 51.50 ± 3.54 | 0.003* |
Highest CK (IU/L) | 7160.65 ± 5629.42 | 10339.50 ± 8150.56 | 0.203 |
Clinical symptoms |
Muscular symptoms | 28 (100.00) | 12 (100.00) | 1.000 |
Muscle weakness | 28 (100.00) | 12 (100.00) | 1.000 |
Myalgia | 7 (25.00) | 0 (0.00) | 0.081 |
Dyspnea | 0 (0.00) | 0 (0.00) | 1.000 |
Dysphagia | 4 (14.29) | 1 (8.33) | 1.000 |
Extra-muscular symptoms | 6 (21.43) | 1 (8.33) | 0.586 |
ILD | 1 (3.57) | 0 (0.00) | 1.000 |
Arthralgia | 0 (0.00) | 0 (0.00) | 1.000 |
Loss of weight | 1 (3.57) | 0 (0.00) | 1.000 |
Skin involvement | 5 (17.88) | 1 (8.33) | 0.772 |
Treatment |
With IVIg | 11/23 (47.83) | 7/8 (87.50) | 0.095 |
CK creatine kinase, ILD interstitial lung disease, IVIg intravenous immunoglobulin, MMT-8 manual muscle testing. * P < 0.05.
Pathological manifestations of LGMD-like (A-C) and non-LGMD-like anti-HMGCR-positive IMNM patients (D-F). LGMD-like patient’s biopsy showed profound myo-fibers with atrophy (black arrowhead) and central nucleus (black arrow) (A) and focal CD4 (black arrow) (B) and CD68 lymphocytic infiltration (black arrow) (C). Non-LGMD-like patient’s biopsy showed predominant myo-fibers with necrosis and regeneration (white arrow) (D) and scattered CD4 lymphocytic infiltration (black arrow) (E) and focal CD68 lymphocytic infiltration (black arrow) (F). Scale bar = 100 µm.